KR19990029792A - Semiconductor integrated circuit device - Google Patents

Abstract

The semiconductor integrated circuit device includes a main memory unit and a sub memory unit including a plurality of memory cell groups, and bidirectional data transfer is performed between an arbitrary region of the main memory unit and each of the plurality of memory cell groups, and the plurality of memory cell groups. Each function as an independent cache memory. Thus, the semiconductor integrated circuit device of the present invention has a main memory suitable for being accessed from a plurality of data processing devices.

Description

Semiconductor integrated circuit device

TECHNICAL FIELD The present invention relates to a semiconductor integrated circuit device, and more particularly, to a semiconductor integrated circuit device having a main memory unit and a sub-memory unit formed in the semiconductor substrate, and a data transfer circuit provided between the main memory unit and the sub memory unit. .

In general, relatively low-cost, low-cost semiconductor devices with atmospheric storage, such as general-purpose DRAMs, are used as main memory of computer systems.

In recent computer systems, the operating speed of the DRAM constituting the main memory has increased with the increase in the system, especially its MPU operating speed. However, the operating speed of DRAM is still not enough, and to solve this problem, it is customary to provide a sub memory between the MPU and the main memory. Generally, such sub memory is referred to as cache memory and consists of high speed SRAM or ECLRAM.

In general, cache memory is provided outside or inside the MPU. In modern workstations or personal computers, a semiconductor memory device is composed of a high speed SRAM, which is a DRAM and a cache memory, which constitute a main memory, formed as one, and the same semiconductor substrate is used. Japanese Unexamined Patent Publications Nos. 57-20983, 60-7690, 62-38590 and 1-146187 disclose examples of such semiconductor memories. Such memory is often referred to as cache DRAM or CDRAM because it includes DRAM and SRAM that function as cache memory and the like. Such a cache memory can transfer data bidirectionally to DRAM and SRAM. These prior arts have problems such as delay of data transfer operation in case of cache misshit, and propose techniques for solving such problems. Examples of the proposed techniques are disclosed in Japanese Patent Laid-Open Nos. 4-252486, 4-318389 and 5-2872. In the techniques disclosed in these patent publications, a latch or register function is provided to the bidirectional data transfer circuit between the DRAM unit and the SRAM unit, so that data transfer from the SRAM unit to the DRAM unit and data transfer from the DRAM unit to the SRAM unit can be performed simultaneously. As a result, the data transfer rate (copy back) may be increased at cache miss hits. This will be described with reference to Japanese Patent Laid-Open No. 4-318389 as an example. 92 schematically shows an example of the configuration of a memory array section of a CDRAM. In FIG. 92, a semiconductor memory device includes a DRAM array 9201 including dynamic memory cells, an SRAM array 9202 including static memory cells, and a bidirectional transfer gate circuit for transferring data between the DRAM 9201 and the SRAM 9202. (9203). The DRAM array 9201 and the SRAM array 9202 are provided with a row decoder and a column decoder, respectively. The addresses given to the row decoder and the column decoder of the DRAM and the row decoder and the column decoder of the SRAM are independently provided through different address pin terminals. 93 and 94 show the configuration of the bidirectional transfer gate circuit in detail. According to this configuration, since the data transfer from the SBL to the GIO and the data transfer from the GIO to the SBL are made through different data transfer paths, performing these data transfers at the same time by the function of the latch 9302 and the amplifier 9306 is possible. It is possible.

However, the CDRAM described above has the following problems. First, since address pin terminals and control terminals are provided separately for the DRAM array and the SRAM array, the number of external pin terminals is much larger than that of a single DRAM. Therefore, there is no compatibility with a substrate having a semiconductor memory mounted thereon with respect to compatibility with a conventional DRAM or the like. Second, in the bidirectional transfer gate circuit, the number of circuits each having an area large enough to realize the above-described transfer is limited, so that the number of transfer buses is limited. As a result, the number of bits that can be transferred between the DRAM array and the SRAM array at one time is limited to 16 bits. In addition, since the transmission bus is arranged in an area where the column selection line is not arranged, the number of transmission buses is limited by the width of the area. In general, the smaller the number of bits transmitted at one time, the lower the cache bit rate.

Japanese Patent Laid-Open No. 5-210974 discloses a technique in which an address input signal pin of a CDRAM is commonly connected to both a DRAM array and an SRAM array. 95 and 96 show the structure of this technique. In this example, as in CDRAM, there is a second problem in which the number of bits transmitted at one time between the DRAM array and the SRAM array is limited to 16 bits. 97 and 98 show a configuration in which the storage capacity of the SRAM is increased to improve the cache hit rate. However, this configuration eliminates substrate compatibility due to input pins for selecting SRAM cells, but solves the second problem of limiting the number of bits transferred at once between a DRAM cell array and an SRAM cell array to 16 bits, as in CDRAM.

Another example in the art is EDRAM, which is DRAM with cache SRAM, as disclosed, for example, in EDN, pages 46-56, issued January 5, 1995. The EDRAM shown in FIG. 99 has a different configuration from general-purpose DRAMs having the same storage capacity, and even though DRAMs and SRAMs share address input terminals in common, they do not have substrate compatibility. The number of bits transmitted to the SRAM at one time is equal to the number of sense amplifiers that are activated at one time. In this example, 512 (× 4) bits are transmitted at one time. Although in this EDRAM configuration, the number of bits to be transferred at one time is large, the SRAM holding data has only one set (one row) of storage capacity for the bits to be transferred at one time. Although typically the larger the number of bits transmitted at a time, the larger the cache hit rate, the EDRAM has only one set (one row) of cache memory, increasing the cache mishit rate, thus increasing the overall system speed. The improvement cannot be achieved sufficiently. In order to increase the number of sets of cache memories in the EDRAM, an additional SRAM register, a block selector, and the like must be additionally provided for each block of the predetermined number of DRAM cell arrays to substantially increase the area occupied by the circuit. do.

In addition, there is a problem of deterioration in the cache hit rate when there is an access request from a plurality of recent processing apparatuses as shown in FIG. When CDRAM or EDRAM is used as the main memory as shown in FIG. 100 and there is an access request from a plurality of processing devices (memory masters), the cache hit rate is reduced and the number of address requests of different sets (rows) can be increased. The speedup of overall system operation is limited.

Background of the Invention With the spread of systems having a plurality of processing apparatuses (memory masters), memory units are required to respond to various types of access requests rather than one type of access request as in the conventional memory portion. In other words, a memory having a configuration different from that of the conventional memory is required.

Accordingly, an object of the present invention is to respond to the main memory unit and the plurality of access requests in order to achieve high-speed operation of the entire system without reducing the cache hit rate even when there are access requests from the plurality of memory masters. A semiconductor integrated circuit device including an assignable sub memory unit is provided.

Another object of the present invention is to provide a semiconductor integrated circuit device having a main memory section and a sub memory section whose external terminals are configured similarly to the configuration of the external terminals of the main memory section.

It is still another object of the present invention to provide a semiconductor integrated circuit device having a main memory and a sub memory, in which the number of bits can be transferred between the main memory section and the sub memory section at one time and the set number thereof can be optimized.

Another object of the present invention is to provide a semiconductor integrated circuit device having a main memory and a sub memory, in which a read / write operation of the sub memory unit and a data transfer operation between the main memory unit and the sub memory unit can be simultaneously performed.

In order to achieve the above object, a semiconductor integrated circuit device according to the present invention includes a main memory unit and a sub memory unit composed of a plurality of memory cell groups, and bidirectionally between an arbitrary region of the main memory unit and each of the plurality of memory cell groups. Data transfer is performed, and the plurality of memory cell groups each function as an independent cache memory.

According to a second aspect of the present invention, a main memory portion of a semiconductor integrated circuit device is composed of a dynamic memory cell array, a static memory cell array, a mask ROM cell array, a ferroelectric memory cell array or a flash memory cell array, and the sub memory portion is a static type. It is composed of a memory cell array.

According to the third aspect of the present invention, the sub memory section of the semiconductor integrated circuit device is composed of a static memory cell having dual ports.

According to a fourth aspect of the present invention, bidirectional data transfer between the arbitrary region of the main memory section and the memory cell group of the sub memory section is performed in units of a predetermined number of bits.

According to the fifth aspect of the present invention, the semiconductor integrated circuit device has the same package pinout as the package pinout of the semiconductor memory device consisting only of the main memory portion.

According to a sixth aspect of the present invention, a semiconductor integrated circuit device includes a semiconductor memory device including a main memory part and a sub memory part including a plurality of memory cell groups, and a control device for the semiconductor memory device. The control device is integrally formed on the same semiconductor substrate.

The above and other objects, features and advantages of the present invention will become more clearly understood from the following detailed description of the present invention with reference to the accompanying drawings.

1 is a block diagram showing an overall configuration of a semiconductor memory device according to a first embodiment of the present invention.

FIG. 2 is a block diagram of a memory system including a plurality of memory masters for requesting connection to a semiconductor memory device and the semiconductor memory device shown in FIG.

FIG. 3 is a block diagram of a memory system including a plurality of memory masters for requesting connection to a semiconductor memory device and the semiconductor memory device shown in FIG.

FIG. 4 is a block diagram of a memory control device in which the semiconductor memory device and the input signal shown in FIG. 1 are integrally formed on the same semiconductor substrate to generate the semiconductor memory device; FIG.

Fig. 5 is a block diagram of a memory control device in which the semiconductor memory device and input signal shown in Fig. 1 are integrally formed on the same semiconductor substrate to generate the semiconductor memory device.

FIG. 6 is a block diagram of a memory control device integrally formed on the same semiconductor substrate and controlling the semiconductor memory device and input signal generated in the semiconductor memory device.

FIG. 7 is a layout view of external terminals of the semiconductor memory device shown in FIG. 1; FIG.

8 is a layout view of external terminals of the semiconductor memory device shown in FIG. 1;

9 is a layout view of external terminals of the semiconductor memory device shown in FIG. 1;

FIG. 10 is a diagram showing a correspondence relationship between various commands for determining an operation function of a semiconductor memory device and an external terminal shown in FIG. 1; FIG.

FIG. 11 is a view showing an external terminal state showing a read command shown in FIG. 10;

FIG. 12 is a view showing an external terminal state showing a write command shown in FIG. 10; FIG.

FIG. 13 is a view showing an external terminal state showing the prefetch command shown in FIG. 10; FIG.

FIG. 14 is a view showing an external terminal state showing a prefetch command with automatic precharge shown in FIG. 10; FIG.

FIG. 15 is a view showing an external terminal state showing a restore command shown in FIG. 10; FIG.

FIG. 16 is a view showing an external terminal state showing a restore command with automatic precharge shown in FIG. 10; FIG.

FIG. 17 is a view showing an external terminal state showing an active command shown in FIG. 10; FIG.

18 is a view showing an external terminal state showing a precharge command with automatic precharge shown in FIG.

FIG. 19 is a view showing an external terminal state showing a full bank precharge command shown in FIG. 10; FIG.

20 is a diagram showing an external terminal state showing a CBR refresh command shown in FIG. 10;

FIG. 21 is a view showing an external terminal state showing a device non-selection command shown in FIG. 10;

FIG. 22 is a view showing an external terminal state indicating a non-operation command shown in FIG. 10;

Fig. 23 is a diagram showing an external terminal state showing the register setting command (1) shown in Fig. 10;

Fig. 24 is a diagram showing an external terminal state showing the register setting command (2) shown in Fig. 10;

Fig. 25 is a diagram showing an external terminal state showing the register setting command shown in Fig. 10;

Fig. 26 is a diagram showing an external terminal state showing a mode register setting command which is a part of the register setting command shown in Fig. 10;

FIG. 27 is a diagram illustrating an address sequence accessed corresponding to a burst length and each lap time in a data input / output mode. FIG.

Fig. 28 is a timing chart of data output having a burst length of 4 and a read latency of 2 at the time of read command input.

Fig. 29 is a timing chart of data output having a burst length of 4 and a write latency of 0 at the time of a write command input.

30 is a flowchart of data and address designation during a read command operation.

Fig. 31 is a flowchart of data and address designation during a write command operation.

32 is a flowchart of data and address designation during prefetch instruction operation.

33 is a flowchart of data and address designation during a restore command operation.

Fig. 34 is a flowchart of data and addressing in the activation command operation.

35 is an array layout schematically illustrating an array arrangement of a semiconductor memory device according to an embodiment of the present invention.

36 is an array layout schematically illustrating an array arrangement of a semiconductor memory device according to an embodiment of the present invention.

37 is an array layout schematically illustrating an array arrangement of a semiconductor memory device according to an embodiment of the present invention.

38 is an array layout schematically illustrating an array arrangement of a semiconductor memory device according to an embodiment of the present invention.

39 is an array layout schematically illustrating an array arrangement of a semiconductor memory device according to an embodiment of the present invention.

40 is an array layout schematically illustrating an array arrangement of a semiconductor memory device according to an embodiment of the present invention.

FIG. 41 is a schematic view showing an entire chip layout of a semiconductor memory device according to an embodiment of the present invention. FIG.

42 is a schematic diagram illustrating an entire chip layout of a semiconductor memory device according to an embodiment of the present invention.

43 is a schematic view showing an entire chip layout of a semiconductor memory device according to an embodiment of the present invention.

44 is a schematic diagram illustrating an entire chip layout of a semiconductor memory device according to an embodiment of the present invention.

45 is a schematic diagram illustrating an entire chip layout of a semiconductor memory device according to an embodiment of the present invention.

46 is a schematic diagram illustrating an entire chip layout of a semiconductor memory device according to an embodiment of the present invention.

47 is a schematic diagram illustrating an entire chip layout of a semiconductor memory device according to an embodiment of the present invention.

48 is a block diagram schematically illustrating a block using a common power supply of the semiconductor memory device according to the embodiment of the present invention.

49 is a block diagram schematically illustrating a block using a common power supply of the semiconductor memory device according to the embodiment of the present invention.

50 is a diagram showing the wiring arrangement in the DRAM array portion of the semiconductor memory device according to the embodiment of the present invention.

51 is a diagram showing the wiring arrangement of a DRAM array unit, a data transfer unit, and an SRAM array unit in the semiconductor memory device according to the embodiment of the present invention.

52 is a diagram showing the wiring arrangement of a DRAM array unit, a data transfer unit, and an SRAM array unit in the semiconductor memory device according to the embodiment of the present invention.

53 is a diagram showing the wiring arrangement of a DRAM array unit, a data bus line, and an SRAM array unit in the semiconductor memory device according to the embodiment of the present invention.

54 is a block diagram of an operation control circuit of the semiconductor memory device shown in FIG. 1;

FIG. 55 is a diagram showing details of the configuration of a DRAM unit and a data transfer circuit shown in FIG. 1; FIG.

FIG. 56 is a diagram showing in detail an example of an array configuration of the DRAM array 110-1 according to the embodiment of the present invention shown in FIG.

FIG. 57 is a diagram showing in detail an example of a connection relationship between a bit line and a transmission bus line for a part of the layout shown in FIG. 56 (corresponding to four pairs of bit lines); FIG.

58 is a circuit diagram showing a data transmission circuit in detail.

FIG. 59 is a diagram showing an example of the configuration for solving the problem of the example shown in FIG. 57;

60 is a block diagram illustrating an example of a DRAM line control circuit.

FIG. 61 is a diagram showing an example of a specific configuration of a DRAM row decoder and a DRAM row control circuit shown in FIG. 55;

62 is a diagram showing an example of a detailed circuit configuration of a DRAM bit line selection circuit.

63 is a diagram showing an example of a detailed circuit configuration of a DRAM bit line selection circuit.

64 is a diagram showing an example of a detailed circuit configuration of a DRAM bit line selection circuit.

65 is a diagram showing an example of a detailed circuit configuration of a DRAM bit line selection circuit.

FIG. 66 is a view showing a relationship between a pair of data transfer bus lines, a DRAM bit line selection circuit, and an SRAM cell in the array layout shown in FIG. 36;

FIG. 67 is a signal waveform diagram showing the operation of each data transfer bus line shown in FIG. 66; FIG.

Fig. 68 is a diagram showing a detailed example of the data input / output terminal and the SRAM unit shown in Fig. 1;

69 is a diagram showing a configuration example of an SRAM memory cell.

70 is a detailed circuit diagram of a flip-flop of the SRAM cell shown in FIG. 69;

FIG. 71 is a diagram showing a detailed example of a connection circuit for connecting the SRAM bit line shown in FIG. 69; FIG.

FIG. 72 is a diagram showing a detailed example of a connection circuit for connecting the SRAM bit line shown in FIG. 69; FIG.

FIG. 73 shows a detailed example of a connection circuit for connecting the SRAM bit line shown in FIG. 69; FIG.

74 is a diagram showing a detailed example of the SRAM row control circuit shown in FIG. 68;

FIG. 75 shows a detailed example of the SRAM column control circuit shown in FIG. 68;

FIG. 76 shows a detailed example of the latch circuit and multiplexer shown in FIG. 75; FIG.

FIG. 77 is a signal waveform diagram showing the operation of the multiplexer shown in FIG. 76;

78 is a block diagram showing a DRAM array, an SRAM row decoder, and a data control circuit shown in FIG.

79 is a signal waveform diagram showing operations of the DRAM array, the SRAM row decoder, and the data control circuit shown in FIG. 78;

80 is a diagram showing an example of a configuration of an SRAM unit and a data input / output terminal;

81 is a view showing a detailed example in the case of a low redundant line in an SRAM array unit.

82 is a diagram showing examples of power supply voltages applied to a DRAM array unit and an SRAM array unit;

83 is a diagram showing examples of power supply voltages applied to a DRAM array unit and an SRAM array unit.

Fig. 84 is a view showing a simulation result of the write time dependence on the SRAM cell with respect to the power supply voltage.

85 shows a detailed example of the configuration of an SRAM array unit which performs a temporary cell transfer function;

FIG. 86 is a signal waveform diagram showing an operation of an SRAM cell when the data of the SRAM cell is read by performing a temporary cell transfer in FIG. 85; FIG.

87 is a signal waveform diagram showing an automatic continuous prefetch transmission function.

88 shows a detailed example of an SRAM line control circuit for implementing a multi-line continuous read / write function;

89 is a diagram showing a detailed example of a read function of a plurality of row continuous read / write functions.

Fig. 90 is a correspondence table showing the states of the read (3) and write (3) commands of the real time mode setting function and respective input terminals.

91 is a signal waveform diagram showing a real time mode setting function.

FIG. 92 is a schematic view showing the configuration of a memory array section of the CDRAM; FIG.

FIG. 93 is a block diagram of a bidirectional transfer gate circuit of the CDRAM shown in FIG. 92;

FIG. 94 is a circuit diagram of the CDRAM bidirectional transfer gate circuit shown in FIG. 92;

95 is a block diagram schematically illustrating a CDRAM.

FIG. 96 is a circuit diagram of an SRAM of the CDRAM shown in FIG. 95;

97 is a block diagram schematically showing the structure of a CDRAM;

FIG. 98 is a circuit diagram of an SRAM of the CDRAM shown in FIG. 97;

99 is a block diagram schematically showing the configuration of an EDRAM;

100 is a block diagram schematically showing the configuration of a memory system having a plurality of processing devices.

※ Explanation of code for main part of drawing

100: semiconductor memory device of the present invention

101: DRAM

102: SRAM part

103: data transmission circuit

110: DRAM array

111: DRAM memory cell

112: sense amplifier

113: DRAM Low Decoder

114: DRAM column decoder

115: DRAM low control circuit

116: DRAM low control circuit

120: SRAM Array

121: SRAM Low Decoder

122: SRAM column control circuit

123: SRAM Column Decoder

124: SRAM low control circuit

131: data transmission selection circuit

132: first data transfer selection circuit

133: second data transmission selection circuit

150: operation control circuit

151: data latch circuit

152: data output buffer

153: data amplifier

160: data control circuit

180a, 180b, 180c: Memory Master

190: mixed semiconductor device

191: memory controller

192: data buffer

303: SRAM bit line control circuit

304: SRAM column selection circuit

307: Lead / Light Amplifier

308: data input / output circuit

309: temporary low selection circuit

311: flip-flop circuit

312: connection circuit

313: connection circuit

315: balancer circuit

316, 317: connection point of the flip-flop circuit

318: balancer circuit

350: SRAM internal low address latch circuit

351: counter circuit

352: multiplexer

390: first column decoder

391: second column decoder

392: first column address buffer

393: first column address buffer

394 data buffer (SRAM cell)

395: first data latch circuit

396: second data latch circuit

410: internal clock generation circuit

420: command decoder

421: input signal buffer

422 command determination circuit

430: control logic

431: DRAM control circuit

432: transmission control circuit

433: SRAM control circuit

440: address control circuit

450: mode register

460: DRAM internal row address latch circuit

470: Multiplexer

480: internal address counter circuit

490 refresh control circuit

495: DRAM internal column address latch circuit

498: transmission bus control circuit

501: decoder circuit

502: blowing logic

503: register

504: multiplexer

505: data input / output mode memory section

506: counter circuit

507: SRAM internal column address latch circuit

508: Multiplexer

509: decoder circuit

601: DRAM array unit

602: SRAM array unit

603: power supply voltage conversion circuit

DMC: Dynamic Memory Cell

N1: memory transistor

C1: memory capacitor

DWL: DRAM Wordline

DBL: DRAM Bitline

DSA: DRAM Sense Amplifier

DSAP: DRAM Sense Amplifier Control Signal

DSAN: Sense Amplifier Control Signal

iCLK: Internal Clock Signal

iA0-iA13: Internal address signal

iADR0-iADR12: DRAM internal low address signal

iAD13: Bank select signal

DBSW: DRAM Bit Selection Circuit

DBS1-DBS4: DRAM Bit Selection Signal

TBL: Data Transmission Bus Line

TSW: Data Transmission Circuit

GTL: Global Data Transmission Bus Line

iADC5-iADC6: DRAM column address signal

TE: Data transmission enable signal

DMB1-DMB16: Memory Cell Block

DRB1-DRB16: DRAM Low Decoder

SAB1-SAB17: (Sense Amplifier + DRAM Bit Line Selection Circuit + Data Transfer Circuit) Block

SWTR: Switching Transistor

ADRL: DRAM Row Address Latch Signal

ADCL: DRAM column address latch signal

N100-N115, N200, N201, N210-215, N230-N235, N 250, N251, N260, N262, N264, N280: N-channel MOS transistor

P100-P103: P Channel MOS Transistors

R100, R101: Resistance

SMC: SRAM Memory Cell

SBL: SRAM Bitline

TWL1-TWL16: SRAM Cell Data Transfer Low Select Signal

SWL1-SWL16: SRMA Cell Lead / Write Low Select Signal

iASR0-iASR3: SRAM Internal Low Address Signal

SSL1-SSL128: SRAM Column Decoder Output Signal

iASC4-iASC10: Internal SRAM Column Address Signal

SIO: data input / output line

GIO: Global Data Input / Output Line

RWL: Lead / Light Busline

SPE: Flip-flop circuit control signal

SNE: Flip-flop circuit control signal

SRWL: Lead Low Select Signal

SRBL: SRAM Lead SRAM Bitline

SWBL: SRAM Write Bitline

ASRL: SRAM internal low address latch signal

ASCL: SRAM internal column address latch signal

CLKUP: Internal Counter Up Signal

SCE: SRAM internal column address latch circuit control signal

SCSL: SRAM column address multiplexer circuit control signal

SRSL: SRAM Row Address Multiplexer Control Signal

SRUP: Internal counter up signal of the counter circuit 351

(1) Basic configuration

The basic configuration of an embodiment of the present invention will be described.

A semiconductor integrated circuit device according to the present invention includes a semiconductor memory device and a control device of the semiconductor memory device. The semiconductor memory device includes a main memory unit and a sub memory unit, and bidirectional data transfer is possible between the main memory and the sub memory. The sub memory section is composed of a plurality of memory cell groups each capable of functioning as an independent cache memory. In the semiconductor memory device of the present invention, the number of control terminals and the number of address terminals can be equal to the number of control terminals and address terminals required to control the main memory unit.

To describe an integrated circuit device of the present invention, taking an embodiment having a synchronous interface having a x8 two-bank configuration including a 64-Mbit DRAM array as the main memory section and a 16-Mbit SRAM array as the sub-memory section. Shall be.

(2) block diagram

1 is a block diagram schematically illustrating a configuration of an entire semiconductor memory device according to an embodiment of the present invention.

In FIG. 1, the semiconductor memory device 100 includes a dynamic RAM (DRAM) portion 101 as a main memory portion, a static RAM (SRAM) portion 102 as a sub memory portion, and a space between the DRAM portion 101 and the SRAM portion 102. A two-way data transmission circuit 103 for data transmission.

The DRAM unit 101 includes a DRAM array 110 having a plurality of dynamic memory cells arranged in row and column matrices, and a DRAM row control circuit for generating a DRAM row select signal and a bank select signal from internal address signals iA0 to iA13. 115, DRAM row decoder 113 for selecting a corresponding row of DRAM array 110 in response to DRAM row select signals iADR0 to iADR12 and bank select signals iAD13, and internal address signals iA5 and iA6. A DRAM column control circuit 116 for generating a DRAM column selection signal from the memory module, and a DRAM column decoder 114 for selecting a corresponding column in response to the DRAM column selection signals iADC5 and iADC6. The DRAM array 110 also includes a memory cell unit 111 and a sense amplifier 112 for detecting and amplifying data stored in the selected DRAM cell. Further, the DRAM array 110 is divided into a plurality of blocks referred to as banks, and in this embodiment, divided into two A and B, one of which is selected by the bank select signal iAD13.

The SRAM unit 102 includes an SRAM array 120 having a plurality of static memory cells arranged in a row and column matrix, an SRAM row control circuit 124 for generating an SRAM row select signal from the internal address signals iA0 to iA3, SRAM row decoder 121 for selecting one cell of the SRAM cell group (in this embodiment, the cell group divided for each row) in response to the SRAM row select signals iASR0 to iASR3, and internal address signals iA0 to iA3 and iA4 to an SRAM column control circuit 122 for generating an SRAM column select signal from iA13, and an SRAM column decoder 123 for selecting a column by the SRAM column select signals iASC4 to iASC10.

In addition, the semiconductor memory device 100 includes an operation control circuit 150 for controlling an operation of the semiconductor memory device in response to an external input signal and a data control circuit 160 for controlling an external data input / output operation. .

Although in this embodiment, DRAM and SRAM are used as the main memory section and the sub memory section, respectively, the present invention is not limited to this. Instead of DRAM, memory such as SRAM, mask ROM, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), flash EEPROM, and ferroelectric memory may be used as the main memory portion. The memory constituting the main memory section is preferably configured such that its own functions and types can be efficiently used. For example, when DRAM is used as the main memory section, conventional DRAM, EDODRAM, DDR synchronous DRAM, DDR synchronous GRAM, SLDRRAM or Rambus DRAM may be used. Further, any RAM can be used as the sub memory section as long as it can be accessed at a higher speed than that of the memory used as the main memory. In the case where the main memory section is composed of a flash EEPROM, it is preferable that the storage capacity of the sub-memory section is at least half the capacity of the device erasing sector of the flash EEPROM.

(3) system

Since the SRAM control circuit 122 is provided in the semiconductor memory device according to the present invention, as described in detail below, the SRAM column control mode can be changed in the SRAM cell group device. This function enables setting of lap time, burst length, and latency for each SRAM cell group (hereinafter referred to as data input / output mode), so that when the SRAM cell group is selected by providing the setting in advance, The data input / output mode of each SRAM cell group is automatically determined in the semiconductor memory device. Thus, control of data from the outside of the semiconductor memory device or control of processing external data for switching the data input / output mode is unnecessary.

The semiconductor memory device of the present invention has a function that, when receiving a plurality of access requests, the semiconductor memory device receives assignments, designations, reassignments, designations, and reassignments in the SRAM cell group apparatus for every access request.

FIG. 2 shows a memory system having a plurality of memory masters for making an access request to the semiconductor memory device 100 shown in FIG.

In Fig. 2, the SRAM cell groups 01, 02, and 03 are assigned to an access request from the memory master 180a, and the SRAM cell group 04 is assigned to an access request from the memory master 180b, and the SRAM cell group 05 , 06, 07 and 08 are specified in the access request from the memory master 180c. The assignment of SRAM cell groups to these access requests can vary and can change at any time. 2, when the data input / output mode requested for the semiconductor memory device by the memory master 180a is different from the data input / output mode requested for the semiconductor memory device by the memory master 180b, Without using a specific control signal, it is possible to continuously perform data input / output operations for the memory master 180a and data input / output operations for the memory master 180b. In order to realize such an operation, the SRAM column control circuit 122 of the semiconductor memory device 100 includes a data input / output mode memory section. This data input / output mode memory section may correspond to the SRAM cell group as shown in FIG. 2 or 1: 1, or may correspond to the plurality of SRAM cell groups as shown in FIG.

4, 5 and 6 are mounted on the same substrate as that of the semiconductor memory device 100 and the semiconductor memory device for controlling and generating an input signal to the semiconductor memory device 100 in accordance with an access request from the memory master. The mixed semiconductor device 190 composed of the memory controller 191 is illustrated.

Such a mixed semiconductor device 190 has a configuration in which input / output operations for all signals are performed through the memory control device shown in FIG. 4, and the semiconductor memory device 100 as shown in FIG. The configuration may be performed directly, or a data input / output operation may be performed by the semiconductor memory device 100 through the data buffer 192 as shown in FIG. 6. However, the present invention is not limited to the mixed semiconductor device 190 having such a configuration. In the mixed semiconductor device 190, the SRAM cell group or groups can be automatically assigned to the access request from each memory master of the system shown in FIG. 2 or 3 in the semiconductor device.

(4) pin arrangement

7 shows an example of pin arrangement of a package of a semiconductor memory device according to the present invention.

The semiconductor memory device shown in FIG. 7 has a 88-bit, 2-bank configuration including a 64-Mbit DRAM array and a 16-Kbit SRAM array, 400 mil × 875 mil with a synchronous interface, and a lead pitch of 0.8 mm. It is housed in a 54-pin TSOP Type II plastic package. This pin configuration and pin number pin arrangement are the same as the pin number and pin arrangement of a conventional 64-Mbit synchronous DRAM. In addition, the number of pins and pin assignments of the x4-bit configuration (Fig. 8), the x16-bit configuration (Fig. 9), the x1-bit configuration, or the x32-bit configuration is the pin number of the corresponding synchronous DRAM regardless of the number of banks. And pin arrangement.

The signal of each pin is defined as follows.

CLK: The clock signal (CLK) is the reference clock commonly used for all other input / output signals. That is, the clock signal CLK determines the take-in timing and the output signal timing of another input signal. The setting / holding time of each external signal is defined with reference to the rising end of the clock CLK.

CKE: The clock enable signal CKE determines whether the subsequent CLK signal is valid. When the CKE signal is high at the rising end of the CLK signal, the CLK signal is determined to be valid, and when the CKE signal is low, the CLK signal is determined to be invalid.

/ CS: The chip select signal (/ CS) determines whether the external input signals (/ RAS, / CAS, / WE) can be accepted. When the signal / CS is low at the leading edge of the signal CLK, the / RAS signal, / CAS signal and / WE signal input at the same timing enter the operation control circuit. When the signal / CS is high at the leading edge of the CLK signal, the / RAS signal, / CAS signal and / WE signal input at the same timing are ignored.

/ RAS, / CAS and / WE: Each control signal (/ RAS, / CAS and / WE) is combined to determine the operation of the semiconductor memory device.

A0 to A13: The address signals A0 to A13 are input to the address control circuit corresponding to the clock signal, and are sent to the DRAM row decoder, the DRAM column decoder, the SRAM row decoder, and the SRAM column decoder to transfer the cells of the DRAM unit and the cells of the SRAM unit. Used to select. This address signal is input to a mode register to be described later in accordance with an internal command signal in order to set the data input / output mode for internal operation. The address signal A13 is also a bank selection signal of the DRAM cell array.

DQM: The data mask signal (DQM) functions to invalidate (mask) data inputs and outputs in bytes.

DQ0 to DQ7: The data signals DQ0 to DQ7 are input / output data signals.

(5) basic operation

The basic operation of the semiconductor memory device according to the present invention will be described.

Note that the number of instructions and data is only an example and any other combination is possible.

10 illustrates examples of various commands for determining a state of an external input control signal and an operation function of a semiconductor memory device according to the present invention. However, it should be noted that any other combination of external input control signals and various commands that determine the operational function of the semiconductor memory device may be used.

10 shows the state of each input control signal at the rising end of the reference clock signal CLK and the operation determined thereby. The sign H represents a logic high level, L represents a logic low level and X represents an arbitrary level. In addition, the input control signal CKEn-1 in FIG. 10 indicates the state of the input control signal CKE in the period of the reference clock immediately before the target reference clock, and the control signal CKE described later for each command is CKEn. -1.

Hereinafter, the commands shown in FIG. 10 will be described in order.

1. (lead command)

The read command is to perform a data read operation from the SRAM cell.

As shown in Fig. 11, the state of each input control signal at the rising end of the external clock signal is CKE = H, / CS = L, / RAS = H, / CAS = L and / WE = H. When this read command is input, addresses A0 to A3 and addresses A4 to A10 are taken as SRAM row select addresses and SRAM column select addresses, respectively. Data at these addresses is output from DQ0 to DQ7 at a time delayed from the input of the read command by the latency.

When DQM = H with the clock set for the read command, DQ0 through DQ7 are masked and not output externally.

Fig. 30 shows the data flow and address signals of internal operation according to this read command. The SRAM cell is selected by row selection of the SRAM row decoder by the internal address signals iA0 to iA3 and column selection of the SRAM column decoder by the SRAM selection signals iASC4 to iASC10 generated from the internal address signals iA4 to iA13. Is selected. Data of the selected SRAM cell is output externally through the data amplifier in the designated data input / output mode.

2. (Write command)

The write command is to perform a data write operation to the SRAM cell.

As shown in Fig. 12, the state of each input control signal at the rising end of the external clock signal is CKE = H, / CS = L, / RAS = H and / CAS = / WE = L. When a write command is input, addresses A0 to A3 are taken as SRAM row select addresses, and addresses A4 to A10 are taken as SRAM column select addresses. As the data is written, data from DQ0 to DQ7 is taken at a time delayed by the latency from the output of the write command.

When the data DQM = H as the data acquisition clock of DQ0 to DQ7, the data output of DQ0 to DQ7 is masked so that it is not taken in internally.

Fig. 31 shows data flow and address signals of internal operation according to this write command. The SRAM cell selects the row of the SRAM row decoder by the SRAM row selection signals iASR0 to iASR3 generated from the internal address signals iA0 to iA3 and the SRAM selection generated from the internal address signals iASR0 to iASR3 and iA4 to iA13. It is selected by the column selection of the SRAM column decoder by the signals iASC4 to iASC10. Write data taken from DQ0 to DQ7 is written to the selected SRAM cell through the write buffer.

As shown in Figs. 30 and 31, the read command and the write command perform read and write regardless of the DRAM unit and the data transfer unit. Thus, these commands can operate even if the data transfer operation between the SRAM cell group and the DRAM portion other than the SRAM row selected for data input / output, and / or the operation in the DRAM portion still continues. On the contrary, even when the read command or the write command is being operated, it is possible to operate by the data transfer operation between the SRAM cell group and the DRAM portion other than the SRAM row selected for data input / output and the operation command in the DRAM.

3. (prefetch)

The prefetch command is to perform data transfer from the DRAM cell group to the SRAM cell group.

As shown in Fig. 13, the state of each input control signal at the rising end of the external clock signal is CKE = H, / CS = L, / RAS = / CAS = H and / WE = L. In addition, A10 = L and A9 = L. When a prefetch command is input, addresses A0 to A3 are taken as SRAM row select addresses, addresses A4 to A6 are taken as SRAM column select addresses, and A13 is taken as bank select addresses of the DRAM array. In this embodiment, the bank A is selected.

Fig. 32 shows the data flow and address signal of the internal operation according to this prefetch command. Of the DRAM cell groups already selected by the activation command to be described later, the SRAM cell in the bank designated by iA13 is selected. In this embodiment, the bank A is selected. The bit lines of the DRAM cell group are designated by addresses iA5 and iA6. The data of the bit line is amplified by the sense amplifier in the activation command, and the data of the selected bit line is transmitted to the data transfer bus line through the data transfer circuit. The cells on the row of the SRAM selected by the addresses iA0 to iA3 stop the holding of previous data, take in data on the data transfer bus line, and hold the data transferred therefrom. Output from the sense amplifier through the data transmission circuit to the data transmission line is stopped after data transmission.

In this embodiment, the number of data transmitted at one time by the prefetch command is 128x8.

4. (Prefetch command with automatic precharge)

The prefetch command with automatic precharge transfers data from the DRAM cell group to the SRAM cell group and automatically precharges the DRAM unit after the data transfer.

As shown in Fig. 14, the state of each input control signal at the rising end of the external clock signal is CKE = H, / CS = L, / RAS = / CAS = H and / WE = L. Moreover, A10 = H and A9 = L. Similar to the prefetch command described above, when a prefetch command with automatic precharging is input, addresses A0 to A3 are taken as SRAM row select addresses, and addresses A5 and A6 as DRAM column select addresses. A13 is taken as the bank selection address of the DRAM array.

The following describes the data flow caused by the address signal of the internal operation and the prefetch command involving automatic precharge. Of the DRAM cell groups already selected by the activation command to be described later, an SRAM cell in the bank designated by iA13 is selected. The bit lines of the DRAM cell group are designated by addresses iA5 and iA6. In the activation command, the data of the bit line is amplified by the sense amplifier, and the data of the selected bit line is transmitted to the data transmission bus line. The cells on the row of the SRAM selected by the addresses iA0 to iA3 stop the holding of the previous data, take the data on the data transfer bus line and hold the data transferred therefrom. The output from the sense amplifier to the data transmission line through the data transmission circuit is stopped after the data transmission. After a predetermined time from the output stop to the data transfer bus line, the word line is deselected, and an internal operation (potential balance of the bit line and sense amplifier) to be described for the precharge command is performed. After a predetermined time from the input of the prefetch command with automatic precharge, the DRAM automatically enters the precharge (non-select) state.

5. (restore command)

This restore command is to perform data transfer from the SRAM cell group to the DRAM cell group. As shown in Fig. 15, this command is a continuous input command that proceeds in accordance with the external clock signals CLK1 and CLK2.

The state of each input control signal at the rising end of the external clock signal shown in Fig. 15 is CKE = H, / CS = L, / RAS = / CAS = H and / WE = L. Moreover, A10 = L and A9 = H. At the rising end of the first external clock signal CLK1, addresses A0 to A3 are taken as the SRAM row select addresses, addresses A5 and A6 are taken as the DRAM column select addresses, and the second external clock signal CLK2 At the rising end, addresses A0 to A12 are taken as bank selection addresses of the DRAM array as the transfer destination. At the rising ends of CLK 1 and CLK2, the address A13 is taken as the bank select address of the DRAM array. This A13 address input by CLK1 and CLK2 must be identical.

Fig. 33 shows the data flow generated by this restore command and the address signal in the internal operation. The internal address signals i1A0 to i1A12 shown in FIG. 33 are internal address data at the first clock CLK1, and the internal address signals i2A0 to i2A12 are internal address data at the second clock CLK2, and are the same internal address signal. The data of the line is shown for each clock. The data of the SRAM cell group selected by the addresses i1A0 to i1A3 generated from the address at the first clock CLK1 is transferred to the data transfer bus line of the bank selected by the address iA13. Then, the data of the data transfer bus line is transferred to the bit line of the DRAM selected by the addresses i1A5 and i1A6. Thereafter, the bit lines of the DRAM are selected by the addresses i2A0 to i2A12, iA13 is selected, and data of the cell group on the selected word line is output to the corresponding bit lines, respectively. A sense amplifier corresponding to the bit line of the DRAM detects and amplifies the data of the DRAM cell group output to each bit line. The sense amplifier corresponding to the bit line selected by the addresses i1A5 and i1A6 detects and amplifies the write data transmitted from the data transmission bus line. Data output from the DRAM to the bit line through the data transfer bus line is stopped after the word line is raised.

In this embodiment, the number of data transmitted at one time is 128 × 8.

6. (Restore command with automatic precharge)

The restore command with automatic precharging is to perform data transfer from the SRAM cell group to the DRAM cell group and automatically perform precharge of the DRAM unit after the data transfer.

As shown in Fig. 16, the states of each input control signal at the rising ends of the external clock signals CLK1 and CLK2 are CKE = H, / CS = L, / RAS = / CAS = H and / WE = L. And A10 = H and A9 = H. At the rising end of the first external clock signal CLK1, addresses A0 to A3 are taken as SRAM row select addresses, addresses A5 and A6 are taken as DRAM column select addresses, and the next second external clock signal CLK2. ), The addresses A0 to A12 are taken as selection addresses of the DRAM array as the transfer destination. At the rising ends of CLK 1 and CLK2, the address A13 is taken as the bank select address of the DRAM array. This A13 address input by CLK1 and CLK2 must be identical.

The data flow generated by this restore with automatic precharge and the address signal of the internal operation will be described below. The data of the SRAM cell group selected by the addresses i1A0 to i1A3 generated from the address at the first clock CLK1 is transferred to the data transfer bus line of the bank selected by the address iA13. Then, the data of the data transfer bus line is transferred to the bit line of the DRAM selected by the addresses i1A5 and i1A6. Thereafter, the word lines of the DRAM are selected by the addresses i2A0 to i2A12 generated by the address at the time of the next clock CLK2, and the data of the cell group on the selected word lines is output to the corresponding bit lines. The sense amplifier corresponding to each bit line detects and amplifies data of the DRAM cell group outputted to the bit line, and the sense amplifier corresponding to the bit line selected by the addresses i1A5 and i1A6 is the write data transmitted from the data transmission bus line. Detect and amplify. The output of the DRAM to the bit line through the data transfer bus line is stopped after the word line is raised. This word line becomes unselected after a predetermined time therefrom, and an internal operation (equilibrium between the bit line and the sense amplifier), which is represented by a precharge command to be described later, is performed. After a predetermined time from the command, the DRAM automatically enters the precharge state.

7. (Activation command)

The activate command is to activate the selected bank from the DRAM array.

As shown in Fig. 17, the state of each input control signal at the rising end of the external clock signal CLK is CKE = H, / CS = / RAS = L, and / CAS = / WE = H. At the input time of this activation command, the address A13 is taken as the bank select address of the DRAM, and the addresses A0 to A12 are taken as the row select address of the DRAM.

Fig. 34 shows the address signal and data flow of internal operation caused by this activation command. In the bank selected by the address iA13, the word lines of the DRAM are selected by the addresses iA0 to iA12. Data of the DRAM cell group on the selected word line is output to the bit line connected thereto, and a sense amplifier corresponding to each bit line detects and amplifies the data of the DRAM cell group output to the bit line. In this embodiment, the number of data transmitted at one time is 512 x 8.

If another word line selection is made for an already active bank, it is necessary to put the bank into a precharge state and input a new activation command.

This command corresponds to the case where the / RAS signal of a normal DRAM goes low.

8. (Precharge order)

The precharge command is to precharge (disable) the selected banks from the DRAM array.

As shown in Fig. 18, the state of each input control signal at the rising end of the external clock signal CLK is CKE = H, / CS = / RAS = L, / CAS = H and / WE = L. When A10 = L and A13 = valid data when inputting the precharge command, the bank designated by the data of the address A13 is precharged (not selected). The bank is selected at the time of inputting the activation command of this precharge, and if there is no activation command input to the bank designated by this precharge command before the latter command input, the precharge command is invalid.

The address signal and data flow of the internal operation caused by the precharge command will be described below.

The bit line potential and sense amplifier potential are balanced by deselecting the word line of the DRAM in which the bank selected by the address iA13 is activated. After the operation of the precharge command is completed, the selected bank is ready to receive the next activation command.

This precharge command corresponds to the case where the RAS signal of a normal DRAM becomes high.

9. (All bank precharge command)

Every bank precharge command precharges (deactivates) all banks in a DRAM array. By this command, the DRAM unit is set to the precharge state, and the activation state of all the banks can be terminated.

As shown in Fig. 19, the state of each input control signal at the rising end of the external clock signal CLK is CKE = H, / CS = / RAS = L, / CAS = H and / WE = L. Moreover, A10 = H.

The address signal and data flow of the internal operation caused by the precharge command will be described below.

The bit line potential and sense amplifier potential are balanced by making all word lines of the selected DRAM unselected. After the operation of this command is completed, all banks will wait to receive the next activation command input.

All of these bank precharge commands correspond to the case where the / RAS signal of a normal DRAM is high.

10. (CBR refresh command)

The CBR refresh command refreshes the cell data of the DRAM data portion. The address signal required for refresh is automatically generated internally.

As shown in Fig. 20, the state of each input control signal at the rising end of the external clock signal CLK is CKE = H, / CS = / RAS = / CAS = L and / WE = H.

The address signal and data flow of the internal operation caused by the CBR refresh command will be described below.

Addresses iA0 to iA12 and iA13 are automatically generated internally. The bank is selected by the internally generated address iA13, and the word lines of the DRAM are selected by the internally generated addresses iA0 to iA12, so that the DRAM cell group on the selected word line corresponds to the corresponding bits. Output each to a line. A sense amplifier corresponding to each bit line detects and amplifies data of the DRAM cell group output on the bit line. The data detected and amplified by the sense amplifier is again written to the DRAM cell group through the bit line. After a predetermined time from the re-write of data, the word line becomes unselected, the bit line and sense amplifier potential are in equilibrium, and the refresh operation is completed.

11. (non-operation command)

The non-operational commands of CKE = H, / CS = L and / RAS = / CAS = / WE = H shown in Fig. 21 are not execution commands.

12. (Device not select command)

The device non-selection command of CKE = H and / CS = H shown in Fig. 22 is not an execution command.

13. (register setting command)

The register setting instruction sets the setting data in the register in various operation modes.

As shown in Figs. 23 and 24, the state of each input control signal at the rising end of the external clock signal CLK is CKE = H, / CS = / RAS = / CAS = / WE = L. When inputting this command, valid data of addresses A0 to A13 is taken as setting data of the operation mode. Input of register settings by this command is required to initialize the device after the power is connected.

Fig. 25 shows the operation by the address data in the register setting command.

Register setting command parts (a), (b), (c) and (d) shown in FIG. 25 are input by one clock shown in FIG. 23, and another register setting command part (d) to be described later is shown in FIG. Input is made by the two clocks shown.

The register setting command (a) shown in Fig. 25 is a test set of refreshes, similar to a normal synchronous DRAM. This address set is selected at the input of A7 = L and A8 = L.

The register setting command (b) shown in FIG. 25 is an unused set. This address set is selected at the input of A7 = L and A8 = H.

The register setting command (c) shown in FIG. 25 is a device test set. This address set is selected at the input of A7 = H and A8 = H.

The register setting command (d) shown in FIG. 25 is a mode register setting set. This address set is selected at the input of A7 = L and A8 = L, and various data input / output modes to be described later are set. The mode register stores the data input / output mode of each SRAM cell group in the sub memory section.

Fig. 26 is a detailed setting item list of mode register settings.

The mode register set (1) command is to switch between the latency mode and the input / output address sequence (lab type). This command is input by one clock of the external clock signal as shown in FIG. This address set is selected when A6 = L, A7 = L and A8 = L.

The latency mode setting is made by the data of A1, A2 and A3 which are input simultaneously, and the input / output sequence (lap type) is set by the data of A0. The latency mode is set to latency = 2 when A1 = L, A2 = H, and A3 = L, otherwise it is either unset or unused. The input / output address sequence (lap type) is set to sequential if A0 = L and to interleave if A0 = H.

The mode register setting (2) command is an address data set for setting the burst length of all selected rows of the SRAM. In order to input the row designation and the burst data length of the SRAM, two of the external clock signals as shown in FIG. It is input continuously over the clock. This address set is selected when A6 = H, A7 = L and A8 = L.

The SRAM cell group is selected by the data of A1, A2 and A3 of the first clock CLK1, and the burst length of the selected cell group is selected by the data of A3, A4 and A5 of the subsequent clock CLK2. The burst length is set to 1 when A3 = L, A4 = L and A5 = L, and set to 2 when A3 = H, A4 = L and A5 = L, and A3 = L, A4 = H and A5 It is set to 3 when = L, and set to 8 when A3 = H, A4 = H and A5 = L, and set to 16 when A3 = L, A4 = L and A5 = H.

Various data input / output modes will be briefly described below.

Burst Length: The burst length represents the number of data continuously input / output by the input of a read command or a write command. The continuous data input / output is performed based on the clock signal. 27 shows the timing of each signal for the data read, where the burst length is four. That is, when a read command is input in CLK0, four data are output in succession in CLK2, CLK3, CLK4 and CLK5. 28 shows timing of respective signals for data write. Since the burst length is 4, when a write command is input in CLK0, four data are taken in succession in CLK0, CLK1, CLK2 and CLK3.

Latency: Latency refers to the waiting time from the input time of the read command or the write command to the time when data input / output becomes possible by the number of clocks. 27 shows timings of signals in reading data. In this embodiment, the latency at the time of data read is two. That is, when a read command is input at CLK0, data output to the DQ terminal is started at CLK2. 28 shows timing of respective signals during data write. In this embodiment, the latency at data write is zero. That is, when a write command is input at CLK0, data taken from the DQ terminal starts at the same time as the input of CLK0.

Lap Type: The lap type (input / output address sequence) determines the address sequence of data input / output when data is continuously input / output for a time corresponding to the set burst length, and includes sequential and interleaved. . Fig. 29 shows the address sequence of data for each sequential and interleaved.

Another operation is a control function by control of the clock enable signal CKE, as in a normal synchronous DRAM.

Hereinafter, an operation part of the semiconductor memory device according to the present invention will be described.

Read operation when there is data specified from the outside in the SRAM section: As shown in Fig. 30, only the data specified by the read command is output to the outside through the data amplifier.

Read in the case where there is no externally specified data in the SRAM section: After the activation command shown in FIG. 34 is completed, the prefetch command shown in FIG. 32 is executed, and the designated data is transferred to the SRAM section. Thereafter, the specified data is output to the outside via the data amplifier by the read command shown in FIG.

Read in the case where there is write data that has not been restored yet and there is no externally designated data in the SRAM section: Write data is transferred to the DRAM section by the restore command shown in FIG. Thereafter, the activation command shown in FIG. 34 and the prefetch command shown in FIG. 32 are executed, and the specified data is transferred to the SRAM unit. Thereafter, the specified data is output to the outside via the data amplifier by the read command shown in FIG.

(6) layout

1. (array layout)

35 is an array layout schematically illustrating an array arrangement of a semiconductor memory device according to an embodiment of the present invention.

In the arrangement of the array arrangement shown in FIG. 35, the entire DRAM array is divided into DRAM array 110-1 and DRAM array 110-2 and SRAM array 120, and an SRAM column decoder 123 is interposed between DRAM arrays. Is provided. Therefore, data is stored between the cell group on the predetermined DRAM selected by the DRAM row decoder 113 adjacent to the DRAM arrays 110-1 and 110-2, and the cell group on the predetermined row of the SRAM selected by the SRAM row decoder 121. It becomes possible to transmit, and a mapping method of a direct mapping system and a set associative system becomes possible.

The data transfer bus lines for data transfer are arranged so that these lines traverse the DRAM array 110-1, the DRAM array 110-2, the SRAM array 120, and the SRAM column decoder 123.

In this embodiment, DRAM array 110-1 and DRAM array 110-2 correspond to banks A and B, respectively.

Fig. 50 (1) is a plan view of the wiring in the DRAM array portion having such a configuration, and Fig. 50 (2) is a hierarchical view in the cross-sectional direction of the wiring in the DRAM array portion.

The data transfer bus line TBL is formed by a wiring layer above the word line DWL, bit line DBL and wiring not used for the sense amplifiers not shown. As in the conventional DRAM, the column select signal does not exist on top of the DRAM cell, and the data transfer bus line TBL is disposed therein. The operation corresponding to the column selection of a normal DRAM is performed by selectively connecting the bit line and the data transfer bus line with the bit line selection switch DBSW. In this embodiment, one pair from four pairs of bit line pairs is selected and connected to the data transfer bus line. The wiring of the bit line selection signal is arranged so as to cross the data transfer bus line and the bit line. The data transfer busline is disposed parallel to the bit line in the upper cell portion of the cell array of the DRAM portion and perpendicular to the word line. In Fig. 50, the connection between the data transfer bus line and the SRAM array is omitted.

36 illustrates another configuration in which the selection circuit 131 is connected between the DRAM array 110-1 and the SRAM array 120 so that the data transfer busline can be selectively connected in addition to the configuration shown in FIG. 35. It is shown. With this arrangement, by disconnecting the data transfer bus line of the DRAM array on the non-selected side with a signal for selecting one of the DRAM arrays 110-1 and 110-2, the charge / discharge current during data transfer It is possible to reduce the data rate and improve the data transfer rate.

51 (1) and 51 (2) are respectively plan views of the wirings of the DRAM array unit having such a configuration and hierarchical diagrams in the cross-sectional direction of the wirings in the DRAM array unit. The data transfer busline divides all DRAM arrays into a first data transfer busline TBLA and a second data transfer busline TBLB, and one of the data transfer buslines TBLA and TBLB is a data transfer selection circuit. 131 is selected. In this figure, the bit line selection switch and the bit line connected to the data transmission line are omitted.

Similar to Fig. 35, in this embodiment, DRAM array 110-1 and DRAM array 110-2 correspond to banks A and B, respectively. Therefore, in this configuration, immediately after data is transferred between the predetermined bank and the SRAM portion, data transfer between the other bank and the SRAM portion is performed, that is, a bank ping-pong operation is performed, and the bank designation is performed. By disconnecting the load of the data transfer bus on one side in the case, it is possible to ensure that the interval of continuous operation during the bank ping-pong operation is not limited by the operating frequency of the data transfer bus line of the DRAM array unit.

It is also possible to provide a data transfer busline with selection circuitry for further dividing the DRAM array and connecting them further. It is also possible to provide a data transfer busline with selection circuitry for further subdividing the SRAM arrays and connecting them.

37 shows another example of an array layout of a semiconductor memory device to which the present invention is applied.

The configuration shown in FIG. 37 differs from the configuration shown in FIG. 35 in that the DRAM array is not divided and the SRAM array and the SRAM column decoder are provided adjacent to both ends of the DRAM array, respectively. With this arrangement, it is possible to increase the speed of the read or write operation by shortening the distance between the SRAM unit and the data control circuit and the distance between the data control circuit and the data input / output terminal DQ. In this example, it is also possible to transfer data between a cell group on a predetermined row of DRAM selected by the DRAM row decoder 113 adjacent to the DRAM array 110 and a cell group on a predetermined row of SRAM selected by the SRAM row decoder. In addition, the mapping method of the direct mapping method and the set association method becomes possible.

The data transfer bus lines for data transfer are arranged so that these lines traverse the DRAM array 110-1, the DRAM array 110, and the SRAM array 120. In this embodiment, the banks A and B exist in a mixed state in the DRAM array 110.

FIG. 38 shows the layout when the DRAM array shown in FIG. 37 is divided. In this layout, the data transfer busline is divided so that the connection between the DRAM array and the SRAM array 120 is made through the data transfer selection circuit 131 using a wiring layer different from the data transfer busline. In this embodiment, the connection line between the data transfer selection circuit 131 and the SRAM array is a global data transfer bus line (GTL).

52 (1) and 52 (2) are a plan view of the wiring in the DRAM array section and the hierarchical view in the cross-sectional direction of the wiring in the DRAM array section, respectively, in this configuration. In FIG. 52, a data transmission bus line is divided into a first data transmission bus line TBLA and a second data transmission bus line TBLB, and one of the data transmission bus lines TBLA and TBLB is a data transmission selection circuit. Selected by 131, it is connected to the global data transfer bus line GTL. Since this global data transfer bus line GTL is connected to the SRAM array 120, it becomes possible to transfer data between DRAM and SRAM. Although in this embodiment, the DRAM array is divided into two, it is also possible to divide the DRAM more. 53 shows an example in which the DRAM array is finely divided. This configuration has the features of the layout shown in FIGS. 36 and 38. In FIG. 53, the DRAM array is divided into four, the selection of the data transfer bus line by the first data transfer selection circuit 132 and the global data transfer bus line (GTL) by the second data transfer selection circuit 133. In FIG. It is possible to transfer the data between the DRAM and the SRAM by selecting and enabling the mapping between the direct mapping method and the set association method. This DRAM array can be further subdivided. In this case, the data transfer bus line and the first data transfer selection circuit are connected in parallel to the global data transfer bus line.

39 shows another example of an array layout of a semiconductor memory device to which the present invention is applied.

In FIG. 39, a wiring layer different from the data transfer bus lines shown in FIGS. 36 and 38 is used. The wirings for connecting the data transfer bus lines arranged on the DRAM array to the SRAM array are formed in wiring layers different from the wiring layers for the data transfer bus lines. In FIG. 39, an SRAM array and an SRAM column decoder are provided adjacent to the DRAM row decoder. Further, in this embodiment, the cell group on the predetermined row of DRAM selected by the DRAM row decoder 113 adjacent to the DRAM array 110 and the predetermined SRAM selected by the SRAM row decoder 121 adjacent to the SRAM array 120 are specified. It is possible to transfer data between cell groups on a row, thereby enabling a direct mapping method and a set associative mapping method.

As shown in Fig. 40, the data transfer busline does not always need to be disposed adjacent to the DRAM row decoder, and as long as the SRAM array, the DRAM array and the SRAM array are capable of mutual data transfer therebetween, There is no restriction on the location of the arrangement.

In the case where bus lines such as data transfer bus lines and global data transfer bus lines are formed with different wiring layers, these bus lines may be polysilicon wiring, polyside wiring, silicide wiring and / or high voltage in addition to the usual metal wiring. By combining the melting point metal wiring and the like, it may be composed of different wiring layers. In this case, their coupling can be optimized taking into account wiring resistance and process constraints. For example, a combination of (a first layer of aluminum wiring and a second layer of aluminum wiring) or a combination of (a first layer of tungsten silicide wiring and a second layer of aluminum wiring) may be considered.

2. (full layout)

Fig. 41 shows the overall chip layout of the embodiment of the semiconductor memory device to which the present invention is applied.

The semiconductor memory device shown in FIG. 41 includes a x8-bit as a main memory section, a 64-Mbit DRAM array in a 2-bank configuration, a 16-Kbit SRAM array as a sub memory section, and a synchronous interface. However, it is also possible to construct the semiconductor memory device in other ways.

As shown in FIG. 41, a cross-sectional region is provided that includes a vertical center and a side center. DRAM arrays 110-1, 110-2, 110-3, and 110-4 each having a storage capacity of 16 Mbit are arranged in 45 areas divided by the cross-sectional shape area. The total storage capacity of the DRAM arrays 110-1, 110-2, 110-3 and 110-4 is 64 Mbits. The DRAM row decoder 113 is provided adjacent to the bottom of the DRAM array 110-1 and 110-2, respectively. Similarly, DRAM row decoder 113 is provided adjacent to the top of DRAM arrays 110-3 and 110-4, respectively. SRAM array 120-1, SRAM row decoder 121, and SRAM column decoder 123 are disposed between DRAM arrays 110-1 and 110-2. Similarly, SRAM array 120-2, SRAM row decoder 121, and SRAM column decoder 123 are disposed between DRAM arrays 110-3 and 110-4. The data transfer busline traverses the DRAM array 110-1, the SRAM array 120, and the DRAM array 110-2 to enable data transfer between the selected DRAM cell group and the selected SRAM cell group. Similarly, the data transfer busline traverses the DRAM array 110-3, the SRAM array 120, and the DRAM array 110-4 to enable data transfer between the selected DRAM cell group and the selected SRAM cell group.

The operation control circuit, the data control circuit and the like are disposed in the other parts of FIG. Although other arrangements are possible, in the illustrated embodiment, the input and output terminals are arranged in the central portion of the side.

In Fig. 41, since the main memory section has a two-bank configuration having a layout in which simultaneously selected sections are not concentrated in the sections, when the bank A is selected, the DRAM arrays 110-1 and 110-4 are selected. When the bank B is selected at the same time, the DRAM arrays 110-2 and 110-3 are selected at the same time. That is, as shown in FIG. 48, the arrangement of the DRAM array is configured such that loads such as the internal power supply wiring VCC and the internal ground wiring GND are not biased to one side. Of course, in order to increase the number of divisions so as to disperse the areas to be simultaneously selected or to reduce the areas to be simultaneously selected, the DRAM arrays 110-1 and 110-3 as the banks A and DRAM arrays as the banks B may be used. It is also possible to use (110-2 and 110-4).

42 shows the overall chip layout of another embodiment of a semiconductor memory device to which the present invention is applied.

DRAM arrays 110-1, 110-2, 110-3, and 110-4 are disposed in four divided regions, respectively. The DRAM arrays 110-1, 110-2, 110-3, and 110-4, each having 16 Mbits of storage and grouped into banks A and B, provide a total of 64 Mbits of storage. The DRAM row decoder 113 is provided adjacent to the bottom of the DRAM array 110-1 and 110-2, respectively. Similarly, DRAM row decoder 113 is provided adjacent to the top of DRAM arrays 110-3 and 110-4, respectively. SRAM arrays 120-1, 120-2, 120-3 and 120-4, SRAM row decoder 121 and SRAM between DRAMs 110-1 and 110-2 and between DRAMs 110-3 and 110-4, respectively. The column decoder 123 is disposed. Although FIG. 42 shows the SRAM column decoder 123 as one block for both ends of the SRAM array, it is also possible to provide one SRAM column decoder 123 for each SRAM array. A data transfer busline for exchanging data between a selected DRAM cell group and a selected SRAM cell group traverses them to enable data transfer between the DRAM array 110-1 and the SRAM array 120-1. Similarly, data transfer buslines are provided between other DRAM arrays and other SRAM arrays.

An operation control circuit, a data control circuit, and the like are disposed in the other parts of FIG. Although in the embodiment shown, other arrangements are possible, the input / output signal terminals may be arranged in the central portion of the side.

Fig. 43 shows the overall chip layout of another embodiment of the semiconductor memory device applied to the present invention.

DRAM arrays 110-1, 110-2, 110-3, and 110-4 are disposed in four divided regions, respectively. The DRAM arrays 110-1, 110-2, 110-3, and 110-4, each having 16 Mbits of storage and grouped into banks A and B, provide a total of 64 Mbits of storage. The DRAM row decoder 113 is provided adjacent to the top or bottom of the DRAM array 110-1 and 110-2, respectively. Similarly, an SRAM column decoder 123, an SRAM array 120, and an SRAM row decoder 121 corresponding to each DRAM array 110 are disposed adjacent to the DRAM row decoder 113. A data transfer busline for exchanging data between the selected DRAM cell group and the selected SRAM cell group crosses the DRAM array 110-1 and is connected to the SRAM array through a wiring layer different from the wiring layer of the data transfer bus line.

The operation control circuit, the data control circuit and the like are disposed in the other parts of FIG.

Figure 44 shows the overall chip layout of another embodiment of a semiconductor memory device applied to the present invention.

In FIG. 44, the SRAM array, SRAM row decoder and SRAM column decoder shown in FIG. 43 are changed. However, this arrangement may be further changed as far as possible to transfer data between the selected DRAM cell group and the selected SRAM cell group.

45 shows the overall chip layout of another embodiment of a semiconductor memory device applied to the present invention.

In FIG. 45, the two layout configurations shown in FIG. 41 are combined by dividing the main memory section and the sub memory section more finely. This layout may be configured by combining a plurality of configurations respectively shown in FIG. 41 or using a combination of the configurations respectively shown in FIG. 42.

In addition, the main memory section shown in Fig. 45 is a two-bank configuration in which the selected portions are not concentrated in the portions of the banks A and B at the same time. Therefore, as shown in FIG. 49, the arrangement of the DRAM array is configured such that loads such as the internal power supply wiring VCC and the internal ground wiring GND are not biased to one side. Of course, it is possible to disperse or reduce the area to be selected at the same time.

46 shows the overall chip layout of another embodiment of a semiconductor memory device applied to the present invention.

In FIG. 46, the data transmission busline extends vertically as opposed to FIG. Although the DRAM row decoder and the SRAM row decoder are shown in FIG. 46 as one block for the DRAM array and the SRAM array, a DRAM row decoder can be provided for each DRAM array, and an SRAM row decoder is also provided for each SRAM array. May be

Also, banks on both sides of the DRAM row decoder may be connected via a common data bus line, as shown in FIG.

(7) Detailed description of each block

Hereinafter, each circuit block of the entire block diagram shown in FIG. 1 will be described in more detail.

The following description is for the examples, it should be noted that the present invention is not limited thereto.

1. (operation control circuit)

54 is a circuit diagram of an operation control circuit 150.

The operation control circuit 150 includes an internal clock generation circuit 410, a command decoder 420, a control logic 430, an address control circuit 440, and a mode register 450.

The internal clock generation circuit 410 generates the internal clock signal iCLK from the external input signals CLK and CKE.

The internal clock signal iCLK is provided to the command decoder 420, the control logic 430, the address control circuit 440, and the data control circuit to control the timing of each part.

The command decoder 420 includes a buffer 421 and a command determination circuit 422 for receiving each input signal. In synchronization with the internal clock signal iCLK, the / CS signal, the / RAS signal, the / CAS signal, and the / WE signal are transmitted to the command determination circuit 421 to generate the internal command signal iCOM. As shown in Fig. 10, the command generation circuit 421 responds to each input signal in a manner shown in the input terminal state and the command correspondence table.

The control logic 430 generates control signals required to perform the operation specified in these signals in response to the internal command signal iCOM, the internal clock signal iCLK and the register signal iREG. The control logic 430 includes a DRAM sub control circuit 431, a transmission control circuit 432, and an SRAM sub control circuit 433, which generate individual control signals.

The register 450 functions to hold data defined by a combination of data of a specific address input when receiving a specific register write command from the command determination circuit and holding it until the register write signal is input again. The data held in that register is referenced when the control logic 430 is operating.

2. (DRAM section)

(DRAM section and data transmission circuit)

FIG. 55 shows the detailed configuration of the data transfer circuit and DRAM section shown in FIG.

In FIG. 55, the DRAM unit 101 includes a plurality of dynamic memory cells (DMCs) arranged in a matrix. Each memory cell DMC includes a memory transistor N1 and a memory capacitor C1. A constant potential Vgg (1/2 Vcc, etc.) is applied to one terminal of the memory capacitor C1. The DRAM unit also includes a DRAM word line DWL connected to a row of the DRAM cell DMC and a DRAM bit line DBL connected to a column of the DRAM cell DMC. Each bit line is formed in pairs with a compensation bit line. The DRAM cell DMC is located at each intersection of the word line DWL and the bit line DBL. The DRAM unit 101 includes a DRAM sense amplifier DSA corresponding to the bit line DBL. This sense amplifier (DSA) functions to detect and amplify the potential difference between the paired bit lines and is controlled by sense control signals (DSAP and DSAN). The DRAM array has a × 8-bit, 2-bank configuration and 64 Mbits of storage capacity, so the DRAM array has a word line (DWL1 to DWL8192), bit lines (DBL1 to DBL512) and sense amplifiers (DSA1 to DSA512). Include. This configuration corresponds to 1 bank of 1 bit.

The DRAM section 101 includes a DRAM row decoder 113 for selecting one of the word lines DWL1 to DWL8192 and a DRAM row for generating the DRAM internal row address signals iADR0 to iADR12 and the bank selection signal iAD13. Control circuit 115. In addition, the DRAM unit 101 selects one of the pairs of four bit lines by the DRAM bit line selection signals DBS1 to DBS4 generated by the DRAM column decoder 114 and transmits the data through the data transfer circuit 103. DRAM bit line selection circuit DBSW connected to the data transfer bus line TBL. In addition, the DRAM unit 101 includes a DRAM column control circuit 116 for generating DRAM column address signals iADC5 and iADC6 used by the DRAM column decoder.

FIG. 56 shows an example of a detailed array configuration of the DRAM array 110-1 in the overall layout shown in FIG. 41.

In FIG. 56, the DRAM array is divided into 16 memory cell blocks DMB1 to DMB16. DRAM row decoders DRB1 to DRB16 corresponding to each of the memory cell blocks DMB1 to DMB16 and blocks SAB1 to SAB17 corresponding to (sense amplifier + DRAM bit line selection circuit + data transfer circuit) are provided. In this figure, each memory cell block DMB1 to DMB16 has a storage capacity of 1 Mbit in 512 rows x 2048 columns. Note that the number of memory cell blocks is not limited to sixteen.

As shown in FIG. 56, when the DRAM memory cell array is divided into a plurality of parts, the length of each bit line is shortened. Therefore, the capacity of the bit lines is reduced, and when the data is read out, it is possible to increase the potential difference between the paired bit lines. In addition, during the operation, only the sense amplifier corresponding to the memory block including the word line selected by the row decoder is operated, so that it is possible to reduce power consumption due to charging and discharging of the bit line.

FIG. 57 details an example of a connection relationship between a transmission bus line and a bit line in the portion 140 (including four pairs of bit lines) of the layout shown in FIG.

In Fig. 57, the sense amplifier DSA is configured such that the sense amplifier DSA1 corresponding to one column is located at one end of the memory block and the sense amplifier DSA2 corresponding to the subsequent column is located at another end such as the memory cell block. Are arranged zigzag. The zigzag arrangement of the sense amplifier is such that the miniaturization of the sense amplifier is not achieved in proportion to the miniaturization of the memory cell even if the size of the memory cell is miniaturized by recent process technology. Since it is necessary when there is no margin, it is adopted. Thus, if the bit line pitch is large enough, the sense amplifier may be placed only at one end of the memory cell block. One sense amplifier (DSA) is commonly used by two memory cell blocks through a shared selection circuit. Each bit line pair is provided with a bit line control circuit for balancing and precharging the potential difference between the bit lines. This bit line control circuit may be used by two memory cell blocks similarly to a sense amplifier.

The bit lines and data transfer bus lines are used for switching the DRAM bit line selection circuits DBSW1 to DBSW4 selected by the DRAM bit line selection signals DSB1 to DSB4 and the respective transistors SWTR shown in detail in FIG. 58. It is connected via the circuits TSW1 and TSW2. The data transfer activation signals TE1 and TE2 for activating this data transfer circuit are obtained by the logical operation of the transfer control signal generated by the operation control circuit shown in Fig. 54 and the address signal for selecting the memory cell block. In Fig. 57, since the bit line is connected to the data transfer bus line through the DRAM data transfer circuit and the data transfer circuit, the data transfer circuit of the memory cell block, which is not activated, becomes non-conductive and selects the DRAM bit line connected thereto. There is no load on the circuit. Thus, it is possible to minimize the load on the data transmission busline during operation. However, the configuration shown in Fig. 57 has a problem in that the chip area is increased because a data transfer circuit is arranged and a wiring for transmitting the data transfer activation signal for the data transfer circuit is provided. 59 shows an example of a configuration for solving such a problem.

In Fig. 59, the bit line and the data transfer bus line are connected to each other only through the DRAM bit line selection circuits DBSW1 to DBSW4 selected by the DRAM bit line selection signals DBS1 to DBS4. This configuration can be realized by providing the function of the data transfer circuit to add logic of the data transfer activation signal to the DRAM column decoder which generates the DRAM bit line selection signals DBS1 to DBS4. Although the load on the data transfer busline is reduced during operation, it is possible to make the chip area very small.

The activation of the DRAM section, column selection, and data transfer operations will now be described with reference to FIGS. 55 and 57.

First, the activation of the DRAM unit will be described. In FIG. 55, the DRAM row select signal and the internal address signals iA0 to iA13, which are one of the DRAM sub control signals generated by the operation control circuit shown in FIG. 54, are input to the DRAM row control circuit 115, The bank select signal IAD13 and the internal DRAM row address signals IADR0 to IADR12 are generated, and the word line DWL of the bank designated by the DRAM row decoder 113 is selected. With the selected word line DWL, data held in the cell DMC is input to the bit line DBL. The potential difference between the paired bit lines is detected and amplified by the sense amplifier in response to the sense amplifier drive signals DSAN and DSAP. The number of sense amplifiers simultaneously activated in the DRAM section 101 is 512, and since the DRAM addition has an 8-bit configuration, the total number of sense amplifiers simultaneously activated is 512x8 = 4096.

Hereinafter, column selection and data transfer of the DRAM unit will be described.

In the DRAM sub column selection circuit 116 shown in FIG. 55, a control signal which is one of the DRAM sub control signals generated by the operation control circuit shown in FIG. 54 in response to the internal address signals iA5 and iA6 It is inputted to generate DRAM column address signals IADC5 and IADC6. The DRAM column address signals IAC5 and IADC6 are input to the DRAM column decoder 114, and after the DRAM bit line selection signals DBS1 to DBS4 are generated, the bit lines are selected, and the data of the bit lines are shown in FIG. The data is transferred to the data transfer bus line TBL by the data transfer activation signal TE obtained according to the logic between the transfer control signal generated by the operation control circuit shown in Fig. 2 and the address signal for selecting the memory cell block. As shown in Fig. 59, it is possible to provide a data transfer circuit function in accordance with the logic of the data transfer activation signal in the DRAM column decoder, and the column bit selection and transfer operations of the DRAM bit line selection signals DBS1 to DBS4 are performed simultaneously. It can be a possible signal.

In FIG. 59, assuming that the DRAM bit line selection signal DBS1 is selected, a signal synchronized with the transmission control signal is input to the DRAM bit line selection circuit DBSW1, and the bit amplified by the sense amplifier DSA1. Data on lines DBL1 and / DBL1 are transferred to data transfer bus lines TBL1 and / TBL1. The DRAM section 101 shown in Fig. 55 includes 128 parts shown in Fig. 59, respectively, and has a x8-bit configuration, so that the total number of data transferred simultaneously from the bit line to the data transfer bus line is 128 x 8 = 1024 pieces. The total number of data transmitted simultaneously is the same as in the other bit configurations.

(DRAM Low Control Circuit and DRAM Low Decoder)

60 is a block diagram of the DRAM row control circuit 115.

This DRAM row control circuit 115 includes a DRAM internal row address latch circuit 460, a multiplexer 470, an internal address counter circuit 480, and a refresh control circuit 490.

In the normal activation of the DRAM unit, the DRAM row control circuit 115 supplies the DRAM internal row address signals IAD0 to IADR12 and the bank selection signals IAD13 to the DRAM row address latch signal ADRL and the internal address signals iA0 to. The iA13 is output from the address latch circuit 460 to which the iA13 is input to the DRAM row decoder 113 through the multiplexer 470.

In the refresh operation, the DRAM row control circuit 115 receives the refresh control signal, and the refresh control circuit 490 outputs a selection signal from the internal address counter circuit, such that the internal address counter circuit 480 is multiplexer 470. Operate to control. As a result, the DRAM internal row address signals IADR0 to IADR12 and the bank select signal IAD13 are input to the DRAM row decoder 113 without input of the address signal. The internal address counter circuit 480 automatically subtracts or sums addresses according to a predetermined method every time a refresh operation is performed, thereby enabling automatic selection of all DRAM rows.

(DRAM column control circuit and DRAM column decoder)

FIG. 61 shows a detailed example of the DRAM column decoder and DRAM column control circuit shown in FIG.

In FIG. 61, the DRAM column control circuit 116 includes a DRAM internal column address latch circuit 495, which includes internal address signals iA5 and iA6 and data transfer from the DRAM cell to the SRAM cell (prefetch transfer operation) and the SRAM. The DRAM internal column address signals IADC5 to IADC6 are generated by the DRAM column address latch signal ADCL taken in the clock cycle of the command input of data transfer (restore transfer operation) from the cell to the DRAM cell. This DRAM column address latch signal ADCL is one of transmission control signals generated in the operation control circuit shown in FIG. The DRAM column decoder 114 decodes the DRAM internal column address signals IAC5 to IADC6 generated by the DRAM column control circuit 116 to activate the memory block selection address signal and the transfer control signal T. Outputs the DRAM column select signal generated. Therefore, the activation signals TE1 and TE2 of the data transfer circuit shown in Fig. 57 are output signals of the DRAM column decoder 114, and the operation of the data transfer circuit is performed by the DRAM bit line selection circuit to be described later.

(DRAM bit line selection circuit)

62 to 65 show a detailed example of the circuit configuration of the DRAM bit line selection circuit shown in FIG.

FIG. 62 shows the simplest configuration including N-channel MOS switching transistors N200 and N201 (hereinafter referred to as NMOS transistors), where DRAM bit line DBL and data transfer bus line TBL are selected DRAM columns. It is connected by a signal.

63 shows an NMOS transistor having a gate connected to each pair of DRAM bit lines and differentially amplifying the data on the DRAM bit line when the data on the DRAM bit line DBL is transferred to the data transfer bus line TBL. DRAM bits comprising N210 and N211 and switching transistor circuits having NMOS transistors N212 and N213 for transferring the amplified signal to the data transfer bus line TBL by a prefetch transfer DRAM column selection signal. Another configuration of the line selection circuit is shown. One of the NMOS transistors N210 and N211 is connected to a fixed potential, such as ground potential. When data on the data transfer bus line TBL is transferred to the DRAM bit line DBL, the DRAM bit line selection circuit is a switching transistor composed of the NMOS transistors N214 and N215 as the circuit shown in FIG. And a DRAM bit line DBL and a data transfer bus line TBL are connected by a restore transfer DRAM column select signal.

FIG. 64 shows a DRAM bit line in the case where the data on the DRAM bit line DBL is transferred to the data transfer bus line TBL with the gate connected to each pair of DRAM bit lines as in the configuration shown in FIG. Switching with NMOS transistors N230 and N231 for differentially amplifying the data on the phase and NMOS transistors N232 and N233 for transferring the amplified signal to the data transfer bus line TBL by a prefetch transfer DRAM column selection signal. Another configuration of a DRAM bit line selection circuit including a transistor circuit is shown. One of the NMOS transistors N230 and N231 is connected to a fixed potential, such as ground potential. In addition, when data on the data transfer bus line TBL is transferred to the DRAM bit line DBL, the DRAM bit line selection circuit includes the NMOS transistors N250 and N251 and the restore transfer DRAM as shown in FIG. And a switching transistor circuit composed of NMOS transistors B235 and N235 for transferring the signal amplified by the column select signal to the DRAM bit line. One of the NMOS transistors N250 and N251 is connected to a fixed potential, such as ground potential.

FIG. 65 shows another configuration of the DRAM bit line selection circuit which realizes the configuration shown in FIG. 64 using only one data transfer bus line. In the configuration shown in FIG. 65, the NMOS transistor N260 eliminates the data transfer bus line by the potential of the DRAM bit line without differentially amplifying the data on the DRAM bit line DBL. NMOS transistor 280 operates similarly. Alternatively, it may consist of only the switching transistor circuit, as shown in FIG. 62. By using only one data transfer busline, the wiring layout is simplified, which makes it possible to reduce noise between the data transfer buslines.

63 to 65, in the configuration in which data is received and transmitted by the DRAM bit line or the data transfer bus line, the DRAM bit line can be completely separated from the data transfer bus line. Therefore, almost no noise generated from one side is transmitted to the other side, thereby enabling high-speed operation.

(DRAM Bit Line Selection Circuit and SRAM Cell Configuration)

FIG. 66 is a configuration diagram showing a relationship between a pair of data transfer bus lines, a DRAM bit line selection circuit, and SRAM cells in the layout shown in FIG. 36;

In FIG. 66, memory cells on one column of a DRAM cell array are connected to a data transfer busline via a DRAM bit line selection circuit, thereby enabling data transfer between the cell and cells on one column of the SRAM cell array. The data transfer bus line and the SRAM cell array are connected to each other by the transfer bus control circuit 498. The data transfer bus control circuit 498 is adapted for selecting and connecting one of the DRAM arrays (in this example banks A and B) disposed on both ends of the SRAM cell array so that only one activated bank can be connected. Since a circuit is included, it is possible to realize the improvement of the speed of data transfer by reducing the charge / discharge current and by reducing the load of the data transfer bus line. Further, when data transfer of both banks is alternately performed (bank ping-pong operation), data transfer buslines of one bank can be separated from the other side as shown in FIG. 67, so that data transfer can be simultaneously performed on both banks. It becomes possible to perform, which effectively shortens the data transmission period.

As described above, the number of bits that can be transmitted at one time by the semiconductor memory device according to this embodiment is 1024 bits, and the load of the data transmission bus lines is very large, so that the voltage magnitudes of all the data transmission bus lines reach the power supply voltage level. When rising, the peak current and current consumption become very large. In order to substantially reduce this peak current and current consumption, the voltage magnitude of the data transmission bus line should be at most limited to half of its power supply voltage.

However, when the voltage magnitude of the data transfer busline is small, the data transfer rate is reduced to some extent because the data of the SRAM cell must be amplified by such a small potential difference. In order to make only the data transfer bus line TBLS in the SRAM cell portion completely change, a differential amplifier circuit connected to the gate of the data transfer bus line TBLA or TBLB in the DRAM may be provided for differential amplification. Alternatively, a sense amplifier may be provided for amplifying only the voltage on the data transfer busline TBLS in the DRAM bank, with the data transfer busline TBLA or TBLB in the DRAM bank separated. Alternatively, the transmission bus control circuit 498 may include circuitry for balancing or precharging.

3. (SRAM part)

(SRAM section and data input / output terminal configuration)

FIG. 68 shows an example of the detailed configuration of the SRAM unit shown in FIG. 1 corresponding to one bit of the data input / output terminal DQ1. Although this example has a 16-Kbit, x8-bit configuration, the present invention is not limited to this, and various configurations may be taken in combination with the configuration of the main memory having a similar effect.

In FIG. 68, the SRAM portion includes an SRAM cell (SMC), as shown in FIG. 69, which cells can be any other circuit provided to store static data, flip-flop circuit 311, data. A connection circuit 312 provided on both sides of the flip-flop circuit 311 for connection to the transmission bus line TBL, and a connection circuit also provided on both sides of the flip-flop circuit 311 for connection to the SRAM bitline ( 313), respectively. This SRAM section includes the SRAM cell data transfer row select signals TWL1 to TWL16 for activating the connection circuit 312 during data transfer between the DRAM cell and the SRAM cell and the SRAM cell read / write row select signal during read or write ( SRAM row decoder 121 for generating SWL1 to SWL16, and SRAM internal row address signals iASR0 to iASR3 input to the SRAM row decoder 121 based on the internal address signals iA0 to iA3 and the SRAM sub-control signal. And a SRAM row control circuit 124 for generating < RTI ID = 0.0 > Of course, it is also possible to use the SRAM cell data transfer row select signal TWL and the SRAM cell read / write row select signal SWL in common. The SRAM bit line control circuit 303 for balancing or precharging the bit lines, and the SRAM column select circuit 304 for forming the data input / output line SIO and the SRAM bit line SBL, may be used as the SRAM bit line ( SBL). The SRAM unit has an SRAM column decoder 123 for generating the selection signals SSL1 to SSL128 input to the SRAM column selection circuit 304, and the internal SRAMs based on the internal address signals iA0 to iA13 and the SRAM sub control signal. And an SRAM column control circuit 122 for generating the column address signals iASC4 to iASC10. This SRAM bit line control circuit 303 may include a sense amplifier for detecting and amplifying the level of the SRAM bit line SBL. The data input / output line SIO is also connected to the external data input / output terminal DQ via the data input / output circuit 308 and the read / write amplifier 307. This data input / output line (SIO) can be provided separately for read and write.

Since the data transfer transfer bus line TBL and the read SRAM bit line are provided, it is possible to perform the read operation irrespective of the data transfer operation.

(SRAM cell)

70 (a) to 70 (d) show detailed circuits of the flip-flop circuit 311 of the SRAM cell shown in FIG. 70 (a) shows a flip-flop circuit composed of P-channel MOS transistors P100 and P101 (hereinafter referred to as PMOS transistors), and NMOS transistors N100 and N101, and FIG. 70 (b) shows a resistor R100. And a flip-flop circuit composed of R101 and NMOS transistors N100 and N101, and SRAMs are normally employed at both ends thereof. FIG. 70C shows the PMOS transistor P102, the NMOS transistor N102, and the balancer circuit 315 controlled by the power interruption control signals PE and NE in the configuration shown in FIG. 70 (a). ) Shows a flip-flop circuit composed of an added configuration. One of the transistors P102 and N102 and the balancer circuit 315 may be omitted. Fig. 70 (d) shows a configuration similar to that of a conventional sense amplifier used for DRAM, which includes a plurality of flip-flop circuits each having the configuration shown in Fig. 70 (a), and is connected to each other by a control signal SPE. PMOS transistor P103 for controlling 316, NMOS transistor N103 for controlling connection point 317 by control signal SNE, balancer circuit 318 for balancing the connection points 316 and 317 thereof. And the balancer 315 provided in each of the flip-flop circuits is arranged in the row direction as in the case shown in Fig. 70 (c). The power supply voltage may be an external power supply voltage or an internal power supply voltage generated by the power supply voltage converting circuit. The PMOS transistor P103 for controlling the connection point 316 by the power-blocking PMOS transistor P102 and the control signal SPE may be an NMOS transistor, respectively, wherein the level of the control signals PE and SPE is a power source. It may be an internally generated power supply voltage higher than the power supply voltage generated by the voltage conversion circuit.

By reducing the current flowing through the flip-flop using the flip-flop circuit shown in Fig. 70 (c) or 70 (d), it becomes possible to substantially reduce the noise generated during the transmission operation. In addition, by performing data transfer while balancing the corresponding connection points, stable and high speed transfer operation can be realized.

The transistors constituting the flip-flop circuit are not specified and may be the same transistors used in the DRAM sense amplifier or peripheral circuit.

(Connection circuit for SRAM bit line and data transfer bus line)

71 to 73 show circuit examples of connection circuits for connection to the SRAM bit line SBL.

FIG. 71 shows the simplest configuration including the NMOS switching transistors N104 and N105 and connected to the SRAM bitline SBL by the read / write row select signal SWL.

The configuration shown in Fig. 72 shows NMOS transistors N108 and N109 having gates connected to respective opposite terminals of the flip-flop circuit in order to differentially amplify the signal on the terminal when reading data from the flip-flop circuit, and read-low selection. And a switching transistor circuit composed of NMOS transistors N106 and N107 for transmitting the signal amplified by the signal SRWL to the SRAM bitline SBL. One of the NMOS transistors N108 and N109 is connected to a fixed potential, such as ground potential. In the case of writing data to the flip-flop circuit, the configuration shown in FIG. 72 further includes a switching transistor circuit composed of the NMOS transistors N110 and N111 shown in FIG. 71, which writes the SRAM bit line SBL to the write low select signal ( SWWL) to connect to the flip-flop circuit.

73 shows an NMOS transistor (N108 and N108) having a gate connected to each opposite terminal of the flip-flop circuit in order to differentially amplify the signal on the terminal when reading data from the flip-flop circuit as in the case shown in FIG. N109), and a switching transistor circuit composed of NMOS transistors N106 and N107 for transmitting the signal amplified by the read row select signal SRWL to the SRAM bit line SRBL for SRAM read. One of the NMOS transistors N108 and N109 is connected to a fixed potential, such as ground potential. When writing data to the flip-flop circuit, the configuration shown in FIG. 73 uses the gates connected to the pair of SRAM write bit lines to differentially amplify the signal on the SRAM write bit line (SWBL) when writing the data to the flip flop circuit. The branch includes a switching transistor circuit consisting of NMOS transistors N114 and N115 and NMOS transistors N112 and N113 for transmitting the amplified signal to opposite terminals of the flip-flop circuit by the write row select signal SWWL. do. One of the NMOS transistors N114 and N115 is connected to a fixed potential, such as ground potential.

In the arrangement shown in FIG. 72 or 73 in which the opposite end of the flip-flop circuit or the SRAM bitline (SBL) is connected to the gate of a transistor to transfer data, it is possible to completely separate the opposite terminals of the flip-flop circuit from the SRAM bitline (SBL). It is possible. Therefore, the noise generated from one side is hardly transmitted to the other side, thereby enabling high speed operation.

It is also possible to configure the connection circuit to the data transmission bus line TBL in the same manner as shown in FIG. 71, 72 or 73. FIG.

(SRAM Low Control Circuit)

74 shows the circuit configuration of the SRAM row control circuit shown in FIG. 68 in detail.

In Fig. 74, the SRAM row control circuit is composed of the SRAM internal row address latch circuit 350, which includes an internal address signal iA0 to iA3 and a latch signal ASRL which takes an internal address signal in the clock cycle of the read / write command input. By this, the SRAM internal row address signals iASR0 to iASR3 are generated. This latch signal ASRL is one of the SRAM control signals generated by the operation control circuit shown in FIG.

(SRAM column control circuit)

FIG. 75 shows the configuration of the SRAM column control circuit shown in FIG. 68 in detail.

In Fig. 75, the SRAM internal column address latch circuit 507 for latching the internal address signals iA4 to iA10 by the latch signal ASCL generated by the clock cycle at the time of the read / write command input, and the control signal ( SCE) to take the output of the SRAM internal column address latch circuit 507 and to counter up in a predetermined sequence by an internal counter up signal CLKUP operable during a burst operation to read / write to the SRAM. A counter circuit 506 is included, and the SRAM internal column address signals iASC4 to iASC10 are output through the multiplexer 508 which passes the output of either the latch circuit 507 or the counter circuit 506. This multiplexer 508 selects the output of the latch circuit 507 at the clock cycle at the time of the read / write command input and is controlled by the control signal SCSL so that the SRAM internal column address signal can be output as fast as possible. In addition, the SRAM column control circuit according to the present invention provides a completely different data input / output mode, for example, burst length, data input, for a plurality of SRAM cell groups (in this embodiment, SRAM cell groups divided every row). To set the output address sequence and latency, etc., take the data input / output mode in accordance with the state of the internal address iA0 to iA13 in the instruction cycle described previously (2) Only burst length can be set for each SRAM cell group, but it is possible to set data input / output address sequence, latency, etc. for them), data input / output mode, storing the data input / output mode. And a memory unit 505. The data input / output mode memory section 505 includes the import logic 502 provided to each divided SRAM cell group for generating setting data to be taken in accordance with the states of the internal addresses iA0 to iA13, the above-described mode. Register setting (2) Setting data of the data input / output mode of each SRAM cell group (output of the acquisition logic 502) by the output of the decoder circuit 501 selected by the enable signal CRE generated in the instruction cycle. IASR0 to iASR3 outputted from the SRAM internal row address latch circuit 350 in the read / write command cycle using the register 503 provided to each of the takeout logics for taking in " By selectively controlling, a multiplexer 504 for passing one of the outputs of the register 503, which holds the setting data of the SRAM cell group, is included. , And the predetermined data is decoded by the address (iA0 to iA3). The counter circuit 506 takes in the output of the multiplexer 504 and operates the semiconductor memory device in the data input / output mode set in each SRAM cell group. This data input / output mode memory section 505 must be provided for each data input / output mode to be set. The internal counter up signal CLKUP, enable signal CRE, control signals CSE and SCSL, and latch signal ASCL are SRAM sub-control signals generated by the operation control circuit shown in FIG. Of course, it is also possible to make the latch signal ASRL input to the SRAM internal row address latch circuit 350 equal to the latch signal ASCL input to the SRAM internal column address latch circuit 507.

Mode register setting (2) Instead of setting the data input / output mode memory section 505 performed for each SRAM cell group by the instruction cycle, the same setting data of two or more SRAM cell groups are set at once or shown in FIG. The logic of the addresses A4 and A6 can be set at the time of setting the SRAM row data of the mode register setting (2) instruction. For example, if A4 = L and A5 = L, the data input / output mode is set for every SRAM cell group, and if A4 = H and A5 = L, the data input / output mode is at least 2 bits of SRAM low data. Is set to two SRAM cell groups that ignore. In this way, the data input / output mode memory section 505 can be set in various combinations. In addition, the number of the register logic 502 as well as the register logic 502 is not the same as the number of divided SRAM cell groups, and a set of the blow logic 502 and the register 503 are jointly used for a plurality of SRAM cell groups. It is possible to do The addresses iASR0 to iASR3 are not always signals from the row address latch circuit 350 inside the SRAM, and a separate circuit can be provided.

Further, as shown in FIG. 76, the SRAM internal column address latch circuit 507 and the multiplexer 508 so that the internal address is output immediately after the logic between the internal clock signal iCLK synchronized with the internal address and the external reference clock signal. By constructing it, it is possible to generate an internal address signal at high speed. In Fig. 76, INTAi and / INTAi are address signals from the counter circuit 506, and EXTAi and / EXTAi are address signals generated from the internal address signal iAi. These signals are switched by the control signals SCSL and / SCSL and the burst control signal. SCSL is a control signal, / SCSL is a control signal having a phase out of its control signal (SCSL). 77 shows an example of the operation of this circuit. In this circuit configuration, the delay from the internal clock iCLK to the time when the internal address signal Yi is output corresponds to one stage of the inverter and is minimized. These internal address signals Yi and YiB are output as address pulse signals.

(SRAM column decoder and data control circuit configuration)

78 shows an example of the configuration of the SRAM column decoder 123 and the data control circuit.

This circuit configuration includes a first column decoder 390 and a second column decoder 391, in which an SRAM column select signal iASC is transmitted to these column decoders in turn. In order to operate the first and second column decoders with one address selection signal iASC, the first and second column decoders are connected to the first column address buffer 392 and the second column address buffer 393, respectively. . The selection signal lines SSL from the first and second column decoders are arranged in parallel in the column direction, and also correspond to the first data latch circuit 395 and data input / output connected to the data input / output lines SIO correspondingly. A second data latch circuit 396 connected to the output line SIO is provided.

79 shows the internal operation timing of the SRAM column decoder.

Each column address buffer sequentially performs column decoder selection signals iASC-1 and iASC-2 based on the CLK signal. That is, when column address selection is performed continuously as in burst mode, the first column decoder and the second column decoder operate alternately. Data of the columns SSL-1 and SSL-2 selected by each column decoder are sequentially output to the corresponding data input / output lines STO-1 and STO-2. These data input / output lines operate at cycle times twice the required cycle time, and the first data latch circuit 395 and the second data latch circuit 396 temporarily hold the data. These two sets of data are synchronized before the data output buffer and are output from the data input / output terminal DQ at the required cycle time.

By using the above-described configuration, it is possible to increase the cycle speed of the continuous data input / output and / or the continuous data write without increasing the internal operation cycle speed. It is also suitable for synchronous DRAM of DOUBLE DATA RATE (DDR).

(Other configuration of SRAM section including data input / output terminals)

80 schematically shows another example of the configuration of an SRAM unit including a data input / output terminal in the case of having an 8-bit configuration. In the case of data output from the SRAM, the data of the SRAM cells of the selected column is output to each data input / output line SIO. The data input / output line SIO of the selected row is connected to the global data input / output line GIO, and the data is transmitted to the corresponding data amplifier 153. The data is then transferred to the data input / output terminal DQ via the read / write bus line RWL, data latch circuit 151 and data buffer 152. Since the SRAM addition has an 8-bit configuration, eight sets of data input / output circuits operate simultaneously to output eight data. Data is written to the SRAM cell along the opposite path of the data read path. By using this circuit configuration using the data input / output line (SIO) and the global data input / output line (GIO), the SRAM row selection of all SRAM cells becomes unnecessary, and the load on the SRAM row selection signal line is reduced, High speed data input / output of SRAM cells is enabled. Further, by using such a configuration, the load on the data input / output line SIO does not increase, so that high speed operation can be realized even if the number of rows of SRAM cells increases.

(SRAM column redundant circuit)

81 shows an example of the configuration of an SRAM column redundant circuit for an SRAM cell array corresponding to one of the input / output terminals DQ.

In FIG. 81, the redundant SRAM cell column is disposed on top of the SRAM cell array. A redundant data input / output line extends upwards from the SRAM cell array through an SRAM row select switch and a typical (non-long-term) data input / output line extends downward from an SRAM cell array through another row select switch. Global data input / output lines are located on top of the SRAM array and are connected to redundant data amplifiers and write buffers, and conventional (non-long) global data input / output lines are located below the SRAM array and are typically populated with conventional data amplifiers and writes. Connected to the buffer. Switching from the SRAM cell column to the redundant SRAM cell column is performed by switching global data input / output lines or by switching data amplifiers and write buffers. This configuration enables the switching of the SRAM array to the redundant cell column for all input / output terminals DQ, and the difference in access time to the redundant cell column even when switching to the redundant cell column occurs. It is possible to remove it. Although in this embodiment, redundant SRAM cell columns, data input / output lines and global data input / output lines are arranged on top of the SRAM cell array, but their arrangement is not so limited.

(8) other

1. (Power supply voltage)

(Power supply voltage applied to DRAM and SRAM)

82 shows an example of the configuration of power supply connections to the DRAM array unit and the SRAM array unit.

In FIG. 82, this semiconductor memory device includes a power supply voltage converting circuit 603 that generates an internal power supply voltage VINT based on the external power supply voltage VEXT. The internal power supply voltage VINT is applied to the DRAM array unit 601, and the external power supply voltage VEXT is directly applied to the SRAM array unit 602. In the recent DRAM, minimization of the process is improved, and the breakdown voltage of the memory cell is reduced. In order to solve this problem, it is common to make the power supply voltage of the memory cell array smaller than the external power supply voltage. However, since the driving ability of the transistor deteriorates necessarily with the decrease of the power supply voltage, it becomes an obstacle to the operation speed. In this embodiment, the miniaturization of the SRAM unit is limited compared to the miniaturization of the DRAM array unit, and the high speed operation speed of the SRAM unit is achieved by allowing the external power supply voltage VEXT to be used for the SRAM unit. For example, as shown in FIG. 84, where the write speed of the write data to the SRAM cell is a simulation result of the power supply voltage dependency of the data write time in the SRAM cell, the external power supply voltage (VEXT) = 3.3V and the internal power supply voltage. If (VINT) = 2.5V, it is increased by 41%.

83 shows an example of the configuration of power supply connections to the DRAM array unit and the SRAM array unit.

In FIG. 83, this semiconductor memory device includes a power supply voltage converting circuit 603 for generating the first internal power supply voltage VINT1 and the second internal power supply voltage VINT2 based on the external power supply voltage VEXT. The first internal power supply voltage VINT is applied to the DRAM array unit 601, and the second internal power supply voltage VINT2 is directly applied to the SRAM array unit 602. At this time, by making the second internal power supply voltage VINT2 higher than the first internal power supply voltage VINT1, the same effects as those obtained by the configuration shown in FIG. 82 can be obtained. At this time, two power supply voltage converting circuits 603 may be used to generate the first internal voltage VINT1 and the second internal voltage VINT2, respectively.

The substrate potential related to the power supply voltage may be considered in various cases depending on the type of memory cells constituting the main memory unit and the sub memory unit. For example, when the main memory is composed of dynamic memory cells, the substrate potential may be used for the main memory portion lower than other regions of the substrate, or for the main memory portion, the sub memory portion, and the bidirectional data transfer circuit. This substrate potential can be realized by forming P wells, N wells, and deeper N wells in the P-type substrate.

2. (explanation of other functions)

(Function 1: Copy Transfer)

The semiconductor memory device according to the present invention can have a data transfer function between SRAM memory cells on the same column, for example, between memory cell SMC1 and memory cell SMC16 shown in FIG.

With this arrangement, the cell data of one row of the SRAM cell array can be copied to another row at a higher speed than in the case of transfer from the DRAM cell. This function can also be executed without affecting the data transfer operation for the DRAM.

Hereinafter, a data transfer operation from a cell in one row including the memory cell SMC1 to a cell in one row including the memory cell SMC16 will be described with reference to FIG. 68.

First, the SRAM cell read / write row select signal SWL1 is activated so that data of one row of cells including the memory cell SMC1 is transferred to each SRAM bitline. Thereafter, the SRAM cell read / write row select signal SWL16 is activated to transfer the data of each bit line to one row of cells including the memory cell SMC16, and the cell data is rewritten. Since data is transferred using the SRAM bit line SBL, for example, data transfer between one row of cells including a memory cell SMC2 selected by the SRAM cell data transfer row select signal TWL2 and a DRAM cell. This can be performed using the data transfer busline TBL, regardless of the data transfer from the cells in the row including the memory cell SMC1 to the cells in the row including the memory cell SMC16. All these operations are performed by command input, so a command for designating the SRAM cell group to be transmitted and the SRAM cell group to be transmitted must be added.

(Function 2: Send Temporary Cell)

In the configuration of the SRAM cell array section shown in Fig. 68, when the specified data exists, data is newly read by performing data transfer from another row of DRAM cells, and the data is temporarily transferred to the DRAM (restoration transfer). Operation), by performing data transfer from another row of DRAM (prefetch transfer operation), data written to the SRAM cell is read. The cycle time of data transfer to the DRAM cell is described as tRC, and the time from the data transfer from the DRAM cell to the DAM cell (prefetch dunson operation) to the reading of the SRAM data is described as tRAC. It takes time. However, data can be read at high speed by providing the following operation.

85 shows an example of a detailed configuration of an SRAM array unit which realizes such a function.

In FIG. 85, the configuration except that the configuration shown in FIG. 85 additionally includes a selection circuit 309 for selecting a temporary cell row by one row of the temporary SRAM cell and the control signal TCSL. This configuration is substantially the same as the configuration shown in FIG. 68. The control signal TCSL is one of the transmission control signals generated by the operation control circuit shown in Fig. 54 and is generated when data transmission to the temporary cell is performed. Instead, it is possible to configure an SRAM array or add a plurality of temporary SRAM cell rows so that one of the existing SRAM cell rows can be selected as a temporary SRAM cell row.

Hereinafter, referring to FIG. 86, in FIG. 85, the data of the SRAM cell transfers (copys) one row of memory cell data including the memory cell SMC1 to one row of memory cells including the temporary memory cell SMCD. An operation example in the case of being read by transferring (prefetching) data from the DRAM cell to one row of cells including the memory cell SMC1 will be described.

First, an activation command is input to select a predetermined DRAM cell row having data to be read. Then, when a newly added command (temporary cell copy command) is input for data transfer to the temporary SRAM cell, the control signal TCSL is activated. The SRAM cell read / write row select signal SWL1 is activated by the SRAM row address input at the same time as the command, and data is transferred to the address, so that one row of cell data including the memory cell SMC1 is stored in each SRAM. It is sent on the bit line. Then, the SRAM temporary cell read / write row select signal SWLD is activated by the control signal TCSL so that the data of each bit line is transferred to a temporary cell of one row including the temporary memory cell SMCD, and the cell The data is rewritten. This operation is the same as the operation of copying cell data of a predetermined row in the SRAM cell array described in the item of (Function 1). By this operation, it is possible to temporarily store the cell data to be transferred in the DRAM. Thereafter, a prefetch command is input to perform data transfer (prefetch transfer operation) from the DRAM cell to one row of cells including the memory cell SMC1 to transfer the data to be read to the SRAM memory cell. Thereafter, a read command is input to read data from the SRAM cell. As described above, it is possible to remove the time tRC to read data during the time tRAC. Then, the data transferred to the temporary cell is transferred to the DRAM (temporary cell restore transfer operation).

(Function 3: Simultaneous transmission of multiple rows)

The semiconductor memory device of the present invention can have the function of transferring the same data by simultaneously selecting the SRAM cell groups of a plurality of rows when the cell group data in the selected row of DRAM is transferred to the SRAM unit.

This function can be added by a simple circuit. In Fig. 68, a plurality of SRAM cell data is obtained by adding a control signal generated by a newly added command to execute the function to the SRAM row control circuit 124 and controlling the SRAM internal row address signal with the control signal. The transmission row select signal TWL can be sufficiently activated.

(Function 4: Automatic Continuous Prefetch / Restore Transfer)

Among the DRAM cell data of the selected row, when the data of the DRAM cell group selected by the DRAM cell decoder is transferred to the SRAM unit, the semiconductor memory device does not repeat the data transfer by a plurality of commands, but in-chip predetermined by one command. By continuously repeating the transfer operation at intervals corresponding to the (in-chip) delay time, the total data transfer time can be reduced.

87 shows an example of internal operations related to this function.

In this description, it is assumed that one row of DRAM cells is divided into four DRAM cell groups by a DRAM column decoder. However, the DRAM cells may be divided into any number of DRAM cell groups.

In Fig. 87, when a newly added instruction (prefetch (2) instruction) defining this function is input, four internal counter-up signals are continuously generated at intervals corresponding to a predetermined delay time in the chip. A DRAM column control signal for generating an internal column address signal and an SRAM row control signal for generating an SRAM internal row address signal are respectively provided to the counter circuit. The DRAM column address and the SRAM row address are input simultaneously with the command input, taken in by the initial internal counter up signal, and each address is sequentially countered up by the subsequent internal counter up signal. Each data is transmitted in a cycle of four counterup signals.

Similarly, data of a plurality of SRAM cell groups is transferred to a plurality of DRAM cell groups selected by a DRAM row decoder and a DRAM column decoder, and the semiconductor memory device does not repeat data transfer by a plurality of commands in one command. By repeatedly repeating the transfer operation at intervals corresponding to a predetermined in-chip delay time, each of the plurality of DRAM cell groups can be transferred. This function provides a counter circuit to the DRAM column control circuit for generating the internal column address signal and the SRAM internal control circuit for generating the SRAM internal row address so as to generate the internal counter up signal at an interval corresponding to a predetermined delay time in the chip. By generating continuously, it can be realized similarly to the foregoing function.

(Function 5: Multiple Low Continuous Leads / Writes)

The semiconductor memory device of the present invention may have a function of continuously reading / writing all data of the SRAM cell groups of the plurality of rows at predetermined intervals according to a predetermined sequence for the SRAM cell groups of the plurality of rows with one command. .

By such a function, for example, when cell data of one row of DRAM is held in a plurality of SRAM cell groups, it is possible to continuously read / write all the cell data of one row of DRAM in a predetermined sequence. As a result, the load of the chipset and / or the memory controller for controlling the semiconductor memory is reduced to enable the operation of the chipset along with other SRAM cell groups and DRAM units. It is also possible to provide other effects when this function is used in conjunction with function 4.

88 shows an example of the detailed configuration of an SRAM row control circuit for realizing this function.

In FIG. 88, this SRAM row control circuit receives an output of the SRAM internal row address latch circuit 350 shown in FIG. 74 by the control signal SRE, and an internal counter generated when the SRAM column address becomes the upper address. A counter circuit 351 for countering up an output at a predetermined address sequence with an up signal SRUP, and any one of the outputs of the latch circuit 35 and the counter circuit 351 is provided with a row address latch circuit inside the SRAM. A multiplexer 352 to pass through 350 is added. The multiplexer 352 is controlled by the control signal SRSL to select the output of the latch circuit 350 in one cycle at the time of the read / write command input. The SRAM column control circuit has a function of sequentially moving from the column address taken when the newly added instruction defining this function is input to the counter circuit 506 shown in FIG. 75.

89 shows an example of the internal operation of the read function of this function.

In FIG. 89, when a newly added instruction (lead (2) instruction) defining this function is input, a control signal SRSL is generated so that the output of the SRAM internal row address latch circuit 350 is multiplexed. ) Become the SRAM internal row addresses iASR0 to iASR3, and are taken into the counter circuit 351 by the control signal SRE. Thereafter, the column address is incremented in synchronization with the reference clock CLK, and the counter 351 increments the row address by the internal counterup signal SRUP generated when the counter becomes the highest address. After the counter becomes the most significant address, the multiplexer 352 is controlled by the control signal SRSL so that the output of the counter circuit 351 becomes the SRAM internal row addresses iASR0 to iASR3. Therefore, by sequentially shifting the row and column addresses, all the data of the SRAM cell groups of the plurality of rows can be read continuously.

The internal counter up signal SRUP, control signal SRE, and control signal SRSL are SRAM sub-control signals generated by the operation control circuit shown in FIG.

(Function 6: Real Time Mode Setting)

The semiconductor memory device according to the present invention, such as burst length, data input / output address sequence and latency, and the like simultaneously with a read / write command input when a read / write operation is performed on an SRAM cell by command input, It can have a function to set the data input / output mode.

With this function, different data input / output modes can be specified when such a mode is requested. Therefore, it is possible to substantially reduce the load on the chipset and the semiconductor controller for controlling the semiconductor memory device of the present invention to improve system performance. It is possible.

Fig. 90 shows a table of correspondences between the read (3) and write (3) commands according to this function and the state of each input terminal. The table shown in FIG. 90 differs from the table shown in FIG. 10 in that burst length selection is assigned to address terminals A11, A12, and A13 that are not used at the time of read / write command input. Therefore, according to the state of the 3-bit address terminal, it is possible to select / specify the burst length shown in FIG. 26 simultaneously with the input of the read 3 / write 3 command. Although burst length selection is specified in this embodiment, it is similarly possible to specify data input / output modes such as data input / output address sequence, latency, and the like.

91 shows an example of operation in which this function is used.

In FIG. 91, the data input / output address sequence is set to sequential, the latency is set to 2, and the burst length is set to the address signals A11 to A13 (internal address signals iA11 to 1) upon input of the read (3) command. iA13)). This can be realized by controlling the SRAM internal column addresses iASC4 to iASC10 by controlling the counter circuit in the SRAM column control circuit shown in FIG. 75 in accordance with the setting of the burst length as in the normal SDRAM.

(Function 7: Automatic Restore / Prepatch Transfer)

In addition, the semiconductor memory device according to the present invention may have a function of transferring data from a DRAM cell group to an SRAM cell group and then transferring data from another SRAM cell group to the DRAM cell group.

In order to realize this function, the DRAM row address and DRAM column address shown in FIG. 60 and the DRAM internal column address latch circuit shown in FIG. It is possible to realize that function. In addition, by providing such a latch circuit in each bank, it is possible to alternately connect different banks. Therefore, it is not necessary to specify the DRAM row address and the DRAM column address during the restore transfer operation, and the effective time required for the restore operation is shortened, thereby simplifying the control in the chipset and the memory controller for controlling the semiconductor memory device. The load is reduced and system performance is improved.

Similarly, the semiconductor memory device according to the present invention may have a function of transferring data from another DRAM cell group to an SRAM cell group after data is transferred from the DRAM cell group to the SRAM cell group.

According to the present invention, in a system including a plurality of processing apparatuses (memory masters), a plurality of processing apparatuses (memory masters) can be used to increase the operating speed of the entire system without reducing the cache hit even if there are access requests from the plurality of memory masters. It is possible to obtain a semiconductor integrated circuit device having a main memory section and a sub memory section, which can be designated in response to an access request.

Claims (6)

Main memory section, and It comprises a sub memory unit composed of a plurality of memory cell group, And bidirectional data transfer is performed between an arbitrary region of the main memory section and each of the plurality of memory cell groups, wherein the plurality of memory cell groups each function as an independent cache memory. The method of claim 1, The main memory unit may include a dynamic memory cell array, a static memory cell array, a mask ROM cell array, a ferroelectric memory cell array, or a flash memory cell array, and the sub memory unit may include a static memory cell array. Circuit device. The method of claim 1, And the sub memory unit comprises a static memory cell having a dual port. The method of claim 1, And bidirectional data transfer between the arbitrary region of the main memory section and the memory cell group of the sub memory section is performed in units of a predetermined number of bits. The method of claim 1, And the package pin arrangement of the semiconductor integrated circuit device has the same package pin arrangement as the package pin arrangement of the semiconductor memory device consisting of only the main memory portion. The method of claim 1, A semiconductor memory device including a main memory unit and a sub memory unit including a plurality of memory cell groups, and A control device for the semiconductor memory device, And said semiconductor memory device and said control device are integrally formed on the same semiconductor substrate.